INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.
UMI University Microfilms International A Bell & Howell Information C om pany 300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA 313/761-4700 800/521-0600 Order Number 9411973
The role of encoding and context in schizophrenic thought disorder: An evaluation of Chapman’s theory from an encoding specificity perspective
Hotchkiss, Barbara Anne, Ph.D.
The Ohio State University, 1993
Copyright ©1993 by Hotchkiss, Barbara Anne. All rights reserved.
UMI 300 N. ZeebRd. Ann Arbor, MI 48106 THE ROLE OF ENCODING AND CONTEXT IN SCHIZOPHRENIC THOUGHT DISORDER: AN EVALUATION OF CHAPMAN’S THEORY FROM AN ENCODING SPECIFICITY PERSPECTIVE
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By Barbara Anne Hotchkiss, B.S., M.A.
The Ohio State University
1993
Dissertation Committee: Approved By: Herbert L. Mirels {A?— ~ Harvey G. Shulman Co-Adviser Steven J. Beck
Co-Adviser Depqfrtment of Psychology Copyright by
Barbara Anne Hotchkiss
1993 In Memory of
James Thomas Hotchkiss ACKNOWLEDGMENTS
I would first and foremost like to acknowledge my deep appreciation to my
adviser, Dr. Herbert Mirels, whose guidance, devotion, and camaraderie have been
crucial at many stages of my graduate career. The completion of this project would not
have been possible without his enormous support, patience, understanding, and
encouragement. Gratitude is also extended to Dr. Harvey Shulman, who served as an
adviser and made significant contributions to the conceptualization and implementa
tion of my project. In addition, I would like to thank Dr. Steven Beck, who served as a
reader on my Dissertation Committee and consistently provided encouragement during
the completion of the project.
A number of individuals were instrumental in recruiting subjects for this study.
I would like to express my sincere appreciation to Ms. Bobbi Fulton, Mr. Matthew
Stepp, and other staff of Netcare Corporation; Dr. Henry Nasrallah, Dr. Stephen
Olson, and Ms. Mary Lynn of The Ohio State University Schizophrenia Research
Program; Dr. Stephen Stem of The Ohio State University Mood Disorders Clinic; Dr.
Terry Carlson, Dr. Paul Kochanowski, Ms. Sharon Wood, and Ms. Patricia Seeger of the Veterans Administration Outpatient Clinic; Dr. Joseph Walsh and other staff of
North Community Counseling Centers; Ms. Linda Jakes and other staff of Concord
Counseling Services; Ms. Anne Logue, Dr. Marshall Vary, and other staff of Harding
Hospital; Ms. Mary Brett and other staff of Northwest Counseling Services; the staff of Jewish Family Services Employment Resource Center; the late Dr. Sandra Babcock of the Bipolar Bears; and Ms. Mary Virginia Culbertson and others of Recovery. Many of these individuals went to great lengths to assist me, despite busy work schedules.
This project was in part supported by an Ohio State University Graduate
Student Alumni Research Award.
My parents, Corinna and Leonard Hotchkiss, and my sister, Karen Mancone, deserve special recognition for their support of my intellectual pursuits. I would also like to thank my in-laws, Ruta and George Paulson and others of the Paulson clan, who have been a constant source of inspiration.
I would like to express my most heart-felt feelings of gratitude to my husband,
John Paulson, for his understanding of the many long hours of time I spent away from him. His enormous support, nurturing, and tolerance during the lengthy process of completing this project were invaluable. VITA
March 6,1956 ...... Bom, Radford, Virginia
1976-1978...... Research Technician, Department of Psychology, Emory University, Atlanta, Georgia
1979 ...... B.S., Emory University, Atlanta, Georgia
198 0 ...... State of Ohio Trainee in Psychology, Central Ohio Psychiatric Hospital, Columbus, Ohio
1981-1983...... Graduate Teaching Assistant, Department of Psychology, The Ohio State University, Columbus, Ohio
1983...... M.A., The Ohio State University, Columbus, Ohio
1983...... Psychology Trainee, Southwest Forensic Psychiatry Center, Columbus, Ohio
1983-198 4...... State of Ohio Trainee in Psychology, Southwest Community Health Center, Columbus, Ohio
1984-198 5 ...... Psychology Assistant, Reynoldsburg Community Counseling Center, Reynoldsburg, Ohio
1985-198 6 ...... Psychology Intern, Indiana University School of Medicine, Indianapolis, Indiana
1986-198 7 ...... Psychology Technician, V.A. Medical Center, Indianapolis, Indiana
1989...... Graduate Administrative Associate, Graduate School, The Ohio State University, Columbus, Ohio
1992-Present...... Secretary, Board of Directors, Peers Unlimited, Inc., Columbus, Ohio
v 1993 Secretary-Treasurer, Board of Electors, Sigma Xi, Ohio State University Chapter, Columbus, Ohio
PUBLICATIONS Millard, B. A. (1983). Hypothesis testing theory. Some boundary conditions. Unpublished master’s thesis, The Ohio State University, Columbus, Ohio.
Newlin, D. B., Hotchkiss, B., Cox, W. M., Rauscher, F., & Li, T.-K. (1989). Autonomic and subjective responses to alcohol stimuli with appropriate control stimuli. Addictive Behaviors. 14,625-630.
Nowicki, S., Jr., Winograd, E., & Millard, B. A. (1979). Memory for faces: A social learning analysis. Journal of Research in Personality. 2,460-468.
FIELDS OF STUDY
Major Field: Psychology
Studies in Clinical Psychology TABLE OF CONTENTS
DEDICATION...... ;...... ii
ACKNOWLEDGMENTS...... iii
VITA...... v
LIST OF TABLES...... ix
LIST OF FIGURES...... xi
CHAPTER PAGE
I. INTRODUCTION...... I
An Overview of Chapman’s Theory of Schizophrenic Thought Disorder...... 6 Seminal Work on the Chapman Theory...... 9 Subsequent Work on the Chapman Theory...... 19 Chapman’s Theory Revisited: Further Considerations...... 57
II. METHOD...... 66
Subjects ...... 66 Thought Disorder Ratings...... 74 Memory Tests...... 75 Procedure...... 79
III. RESULTS...... 86
Thought Disorder Ratings...... 86 Memory Task Performance...... 90 Relationship Between Thought Disorder, Diagnosis, and Memory Task Performance...... 106 Chronicity and Memory Performance in Schizophrenic Subjects 118 Relationship Between Medication and Memory Performance...... 120
IV. DISCUSSION...... 122
Thought Disorder Ratings...... 122 Memory Task Performance...... 124
vii Table of Contents (Continued)
END NOTES ...... 143
APPENDICES
A. Sources of Patient Subjects ...... 145 B. Screening Instruments Used to Determine Eligibility for Participation...... 147
C. Word Triplets Used in the Construction of the Memory Task Materials ...... 150
D. Study Item Lists and Memory Test Answer Sheets for the Memory Tasks...... 156 E. Sample Consent Form...... 208
F. Memory Task Instructions...... 211
G. List Orders of the Memory Tasks...... 217
REFERENCES...... 219
viii LIST OF TABLES
Table Page
1. Characteristics of the subject groups...... 72
2. Format of the memory tasks with examples...... 76
3. Interrater consistency in judgments of thought disorder...... 87
4. Percentage of schizophrenic, bipolar, and normal subjects exhibiting thought disorder symptoms...... 89
5. Mean recall scores and standard deviations for schizophrenic, bipolar, and normal subjects as a function of trial and cue condition at study and test...... 91
6. Analysis of variance of number of words recalled...... 92
7. Mean d1 scores and standard deviations for schizophrenic, bipolar, and normal subjects as a function of trial and cue condition at study and test...... 101
8. Analysis of variance of d' scores...... 102
9. Hierarchical multiple regression of Recall Difference Score 1...... 110
10. Hierarchical multiple regression of Recall Difference Score 2...... I l l
11. Hierarchical multiple regression of Recall Difference Score 3...... 112
12. Hierarchical multiple regression of Recognition Difference Score 1...... 114
13. Hierarchical multiple regression of Recognition Difference Score 2 ...... 115
14. Hierarchical multiple regression of Recognition Difference Score 3 ...... 116
15. Correlations of chronicity measures with memory performance scores in schizophrenic subjects...... 119
16. Correlations of medication measures with memory performance scores in schizophrenic and bipolar subjects...... 121
17. Facilities and programs from which patient subjects were draw n...... 146
ix List of Tables (Continued)
18. Practice list triplets...... 151
19. List 1 triplets...... 152
20. List 2 triplets...... 153 21. List 3 triplets...... 154
22. List 4 triplets...... 155
23. Latin square used in the derivation of the four list orders...... 218
x LIST OF FIGURES
Figure Page
1. Mean number of words recalled by schizophrenic, bipolar and normal subjects as a function of cue condition at study and test...... 94
2. Predicted memory performance profile for the normal group...... 95
3. Predicted memory performance profile for the schizophrenic group...... 97
4. Mean d1 score of schizophrenic, bipolar, and normal subjects as a function of cue condition at study and test...... 104
xi CHAPTER I
INTRODUCTION
Many individuals who are diagnosed with schizophrenia display gross abnormalities in speech and communication. The following excerpt from a psychiatric interview illustrates one type of abnormality that is frequently observed in the disorder.
Interviewer: “What did you think of the whole Watergate affair?”
Patient: “You know I didn’t tune in on that, I felt so bad about it. I said, boy, I’m not going to know what’s going on in this. But it seemed to get so murky, and everybody’s reports were so negative. Huh, I thought, I don’t want any part of this, and I was I don’t care who was in on it, and all I could figure out was Artie had something to do with it. Artie was trying to flush the bathroom toilet of the White House or something. She was trying to do something fairly simple. The tour guests stuck or something. She got blamed because of the water overflowed, went down in the basement, down, to the kitchen. They had a, they were going to have to repaint and restore the White House room, the enormous living room. And then it was at this reunion they were having. And it’s just such a mess and I just thought, well, I’m just going to pretend like I don’t even know what’s going on. So I came downstairs and ’cause I pretended like I didn’t know what was going on, I slipped on the floor of the kitchen, cracking my toe, when I was teaching some kids how to do some double dives.” (Andreasen, 1979a, p. 1319)
To a listener, the verbalizations of schizophrenic individuals may sound confused, bizarre, and nonsensical. In their communication, schizophrenic persons at times may have difficulty adhering to a coherent theme, may ramble on and on without adding to the substance of their talk, may include innumerable irrelevant details, and may com mit a variety of other communication errors. The disturbed nature of their communica tion has often led clinicians and researchers alike to infer that schizophrenia is characterized by thought disorder.
1 2
The notion that aberrations in thinking and cognitive functioning are a fundamental aspect of schizophrenia began with Kraepelin’s (1919/1971) formulations concerning the nature of dementia praecox. Kraepelin, often cited as the father of modern psychiatry, characterized dementia praecox as involving intellectual deterioration and described a multitude of cognitive aberrations observed in the disorder. Later Bleuler (1911/1950), in expanding and revising Kraepelin’s work, renamed the disorder “schizophrenia” and posited a single cognitive defect—a disruption in the coherent, goal-directed processes that normally modulate thinking— to account for schizophrenic behavior.
The writings of Kraepelin and Bleuler stimulated considerable interest in the nature of thought disorder in schizophrenia, and in the decades following their work, a host of theories emerged. These early theories were formulated before the develop ment of theories of cognitive processes in normal individuals. Schizophrenia investigators frequently relied on noncognitive theories from the general experimental literature to explain thought disorder in schizophrenia (Neale & Oltmanns, 1980).
Alternatively, some researchers used observations of patient behavior as a starting point for the development of theories. The early theories tended to be global in nature in that they posited a single underlying mechanism as the source of thought disorder in schizophrenia. In finding differences between schizophrenic and normal subjects on some experimental task, researchers generally inferred—often implicitly—that such differences embodied the fundamental defect in schizophrenia and hence, the mechanism responsible for thought disorder. In this tradition, thought disorder in schizophrenia has been attributed to a “concrete attitude” (Bolles & Goldstein, 1938), errors in logical reasoning (Von Domarus, 1944), a loss of an appropriate response set
(Shakow, 1950), a heightened drive and consequent breakdown in the hierarchical order of competing responses (Broen & Storms, 1961), an active attempt to withdraw
from interpersonal contact (Haley, 1959), and many other proposed causes.
The early studies often lacked the methodological refinement that is common
in contemporary schizophrenia research. Much of the early work was plagued by such
problems as a lack of uniform diagnostic procedures, the absence of proper control groups, and a failure to consider the effects of such potentially confounding variables
as age, IQ, and medication on the behavior under study. Perhaps because of these
problems, attempts to replicate many of the early studies met with varied success.
Some investigators responded to replication failures by postulating that previously
observed performance deficits may characterize some subtypes of schizophrenia but
not others. Subsequent research, however, did not in general support this notion. Still
i other studies ran into conceptual problems. For example, Kopfstein and Neale (1972)
showed that various experimental measures of attention, drawn from different
schizophrenia laboratories, are poorly correlated with one another. Such a finding
weakened the popular claim that attentional processes, as measured by a variety of procedures, are disturbed in schizophrenia.
By the early 1970’s, information processing psychology had become a dominant paradigm in the study of human cognition. Disappointed with the failures of the early studies, many schizophrenia researchers began to utilize the well-articulated models and sophisticated methodologies of information processing psychology to study cognitive processes in schizophrenia. At the same time, these workers began to develop and implement specific procedures in their experimental designs that addressed the special problems of studying clinical populations. A new generation of studies subsequently appeared, involving the comparison of the performance of schizophrenic and control subjects on some specific cognitive task drawn from the information processing literature. With these studies, a myriad of mini-theories came into existence, each of which identified some circumscribed process as aberrant in
schizophrenia. In practice, these mini-theories have been relatively unconcerned with
how their identified deficiencies relate to those of other mini-theories or to thought
disorder in schizophrenia. And, with the burgeoning of the mini-theories, interest in
the early global theories of schizophrenic cognition waned considerably.
More recently, the study of thought disorder in schizophrenia has received
renewed attention. In 1986, for example, an entire edition of Schizophrenia Bulletin
(Holzman), which is published by the National Institute of Mental Health, was devoted to the subject of thought disorder in schizophrenia. In that issue, several scales for
measuring thought disorder were presented along with a number of studies docu
menting the course and nature of thought disorder in schizophrenia and other
psychiatric conditions. These and other descriptive studies conducted in recent years
have shown that, contrary to traditional beliefs, thought disorder symptoms are some
times observed in nonschizophrenic (including normal) individuals as well as in
schizophrenic persons (e.g., Andreasen, 1979b; Andreasen & Grove, 1986; Solovay,
Shenton, & Holzman, 1987). Additionally, some individuals who are diagnosed with schizophrenia do not show abnormal levels of formal thought disorder (e.g., Harrow,
Grossman, Silverstein, Meltzer, & Kettering, 1986; Harrow, Silverstein, & Marengo,
1983). Further, thought disorder symptoms can fluctuate over time (Harrow, Marengo,
& McDonald, 1986; Hurt, Holzman, & Davis, 1983; Spohn, Coyne, Larson,
Mittleman, Spray, & Hayes, 1986), and hence, conceptualizations that dichotomize people as thought-disordered or non-thought-disordered oversimplify the actual state of affairs. And although thought disorder is generally viewed as a unitary phenomenon, recent research reveals that certain types of thought disorder seem to characterize schizophrenia-related disorders, whereas other types of thought disorder 5 symptoms are prominent in other psychiatric conditions (e.g., Andreasen & Grove, 1986; Harrow & Marengo, 1986; Holzman, Shenton, & Solovay, 1986).
Such findings stand in sharp contrast to assumptions, made by early
researchers, that thought disorder is pathognomonic of schizophrenia, and that a comparison of schizophrenic versus normal subjects on some experimental task
provides the essential conditions for studying thought disorder. The practice of
studying schizophrenic performance on cognitive tasks, without reference as to how
performance relates to schizophrenic symptomatology, has been increasingly called
into question. Neale, Oltmanns, and Harvey (1985) point out that schizophrenia
researchers often document some information-processing deficit in schizophrenic
patients and then give lip service to the idea that the deficiency underlies thought
disorder. They present two studies, involving schizophrenic subjects and children at risk for schizophrenia, in which thought disorder was unrelated to information- processing deficits. Neale et al. (1985) suggest that researchers design experiments that will allow an examination of the relationship between thought disorder and laboratory-observed deficits.
The present work considers one of the early theories of thought disorder—
Chapman and Chapman’s theory of accentuation of normal biases in schizophrenia
(Chapman & Chapman, 1965,1973a; Chapman, Chapman, & Miller, 1964). Like the other early theories, the Chapman theory was conceived before the rise of cognitive psychology and has not been examined from a cognitive psychology standpoint in any substantial way. Unlike the other early theories, though, the Chapman theory generally fared well in replication attempts during its heyday (Neale & Oltmanns, 1980, p. 119) and hence shows promise towards making a continued contribution to the understand ing of thought disorder. Although the theory is worded in outdated, associationistic language, its main ideas can readily be translated into information-processing terms drawn from the cognitive psychology literature. Thus, the theory lends itself to examination from a cognitive psychology framework. In addition to evaluating the
Chapman theory using experimental techniques that were developed by cognitive psychologists, this study is aimed at relating schizophrenic performance on prescribed cognitive tasks to measures of thought disorder. Finally, whereas schizophrenia studies typically involve the demonstration of a deficit in schizophrenic performance in order to support hypotheses, the present study relies on the facilitation of schizophrenic performance, under certain conditions, for hypothesis confirmation.
An Overview of Chapman’s Theory of Schizophrenic Thought Disorder
Chapman and Chapman (1973a) proposed that thought disorder in schizophre nia stems from an excessive yielding to normal response biases. They argued that aberrations in schizophrenic speech are mediated by normatively high-frequency responses that are incorrect for the context in which they occur. Thus, schizophrenic speech disturbance does not reflect random, senseless behavior, but results from the inappropriate interpretation and use of words for the task at hand. As an example,
Chapman and Chapman (1973a) cite Bleuler’s (1911/1950, p. 26) description of a patient who began a list of the members of her family with “father, son,” but then added “and the Holy Ghost.” The Chapmans note that although the patient’s verbalizations were clearly deviant in the context of the topic at hand, they do follow a familiar sequence.
Chapman and Chapman (1973a) asserted that the concept of an excessive yielding to normal response biases could readily be operationalized across a variety of experimental contexts. On any given laboratory task, schizophrenic and normal subjects would be expected to commit the same types of errors, but schizophrenic patients are predicted to commit more such errors than normal persons. Consider, for example, a multiple-choice test item in which one of the incorrect response options is
associatively related to the test question. The errors that normal individuals make in
responding to such tests often involve the selection of the associatively related
alternatives. According to the Chapman theory, schizophrenic individuals are
predicted to show heightened predisposition towards responding with such associative
alternatives. In a similar fashion, the response biases that influence performance on
other tasks can be identified in operational terms, thus allowing for precise tests of the
theory.
The Chapmans viewed their formulation as a general principle that could
account for the results of a number of studies from their laboratory. They invoked their
theory (Chapman & Chapman, 1973a) to explain schizophrenic behavior in studies of
similarity judgments (Chapman & Chapman, 1965), vocabulary knowledge (Boland &
Chapman, 1971; Rattan & Chapman, 1973), concept classification (Chapman, 1956a,
1956b, 1958,1961), verbal comprehension (Chapman, 1960; Chapman, Chapman, &
Miller, 1964), and evaluative judgements (Miller & Chapman, 1968, cited in Chapman
& Chapman, 1973a, p. 124). Many of these studies were originally conducted for the
purpose of testing other theories of cognitive dysfunction in schizophrenia. In their
1973 treatise on schizophrenic thought disorder, Chapman and Chapman reevaluated
the studies with respect to the types of errors that subjects produced. They
demonstrated that, in each of the studies, the errors that were most frequently committed by schizophrenic subjects were the same as those that were most often produced by normal individuals. Schizophrenic subjects, however, made more such errors than did normal subjects. In other words, schizophrenic individuals showed the
same response biases as normal subjects, but to a more extreme degree.
Chapman and Chapman (1973a) devoted considerable attention to the relationship between their response bias formulation and other early accounts of schizophrenic thought disorder. They maintained that several of these accounts could be conceptually recast as specific instances of their response bias theory. For instance,
Salzinger (1973) proposed an “immediacy hypothesis,” which holds that schizophrenic
individuals are particularly prone to respond to external stimuli that are immediate by
virtue of their strength, proximity, recency, or ability to provoke an unconditioned
response. Such stimulus characteristics are generally the same as those to which
normal individuals are reactive. Salzinger’s hypothesis can be easily construed as a
special case of the theory that schizophrenia involves an accentuation of normal
response biases. Other early formulations of thought disorder, according to Chapman
and Chapman (1973a), provide possible explanations as to the underlying causes of an
excessive yielding to normal response biases. A number of researchers have put forth
attention deficit formulations, which attribute schizophrenic thought disorder to such defective processes as a failure to disattend from irrelevant stimuli (Cromwell &
Dokecki, 1968), a narrowing in the range of attention (Broen, 1966; Venables, 1964), and an inability to filter out inappropriate stimuli (Weckowicz & Blewett, 1959). The notion that schizophrenic patients have difficulty attending to relevant information is consistent with Chapman and Chapman’s (1973a) assertion that schizophrenic individuals appear to disregard contextual cues more than normals. An accentuation of normal responses biases, then, may stem from defects in attentional processes.
The Chapman theory appears to provide an apt explanation of some of the clinical features of schizophrenia. For example, “loose associations,” often seen in schizophrenia, involve speech in which the flow of ideas is disconnected by repeated shifts from one topic to another with which it is very tenuously or vaguely related
(American Psychiatric Association, 1987). The speech sample given at the beginning of this chapter appears as an example of loose associations in the standard American diagnostic guidelines, the Diagnostic and Statistical Manual of Mental Disorders. 9
Third Edition, Revised (DSM-HI-R; American Psychiatric Association, 1987). In this
excerpt, the patient, when asked about the Watergate affair, began to answer
appropriately, but then contaminated his discourse by introducing topics involving
water and by interweaving them with the topic at hand. Other common symptoms of
schizophrenia (e.g. flight of ideas and tangentiality) also involve the presence of ideas
that are recognizably connected but that are ill-suited for the overall theme of the
discourse. Such diagnostic features fit neatly with the assertion that schizophrenic
individuals err by making normatively high-frequency responses that are inappropriate
for the context at hand.
Seminal Work on the Chapman Theory
A comprehensive review of research pertaining to the response bias theory is
beyond the scope of this dissertation.1 Instead, several of the Chapman studies that
deal most directly with the concerns of the present investigation will be examined in
detail below. The next section will consider follow-up work designed to replicate and
extend the Chapmans’ contributions.
Concept Classification
Chapman (1958) predicted that the performance of schizophrenic individuals
in a concept-classification test would be characterized by an excessive tendency to
give associatively related but incorrect responses to the test material. He administered
a card-sorting task to chronic unmedicated schizophrenic subjects and medical patient controls. For each item in the task, the subjects received a guide-word card, which contained three words, and a sorting-word card, which contained a single word.
Subjects were instructed to select the guide word that was a member of the same conceptual class as the sorting word. Each guide-word card included a word that belonged to the same conceptual category as the sorting word, a word that was 10
associatively related to the sorting word, and a word that was unrelated to the sorting
word. For example, one item included GOLD as the sorting word along with the guide words STEEL (correct alternative), FISH (associative distractor), and TYPEWRITER
(irrelevant distractor). Half of the problems contained associative distractors that were
strongly related to their corresponding sorting words, and the other half contained
weakly related associative distractors. The two levels of associative strength were
combined with two levels of problem difficulty, producing four experimental
conditions.
Chapman (1958) observed that, as is typical in schizophrenia research, the
schizophrenic subjects committed more errors than the controls in all experimental
conditions. In addition, for both subject groups, the number of errors increased as a
function of increasing item difficulty and increasing strength of associative distractors.
On items containing strong associative distractors, the errors made by both groups
were almost always associative in nature. However, in comparison to the normal
controls, the schizophrenic subjects showed a disproportionately greater increase in errors with increasing levels of item difficulty and associative strength. Similar results were obtained when the data were analyzed in terms of a chance-corrected associative error score, which was based on the number of associative errors minus the number of irrelevant errors. Chapman and Chapman (1973a) maintained that the findings just described are consistent with the notion that schizophrenia involves an excessive yielding to normal response biases. Chapman’s (1958) finding that normal individuals committed errors predominantly by selecting associative distractors, and that they increased their associative responses as the task items became harder, suggests that the tendency to select associative distractors constitutes a normal response bias. Schizophrenic 11
individuals, in showing a heightened tendency to respond with associative distractors, thus exhibited an accentuation of a normal response bias.
Yo,pabulary Knowledge Rattan and Chapman (1973) argued that the results of the concept classification
study may have been due to artifacts associated with the differential deficit design
employed by Chapman (1958). A differential deficit design is used in schizophrenia
research to identify deficits that underlie the fundamental nature of schizophrenia and to distinguish them from deficits that stem from a nonspecific decrement in perfor
mance. Individuals with schizophrenia tend to perform more poorly than normals on
most laboratory tasks, and hence, the finding of a schizophrenic deficit on any single
task may merely reflect a generalized deficit rather than a basic mechanism of
schizophrenia (Chapman & Chapman, 1973a, 1973b, 1978; Maher, 1974). Differential
deficit designs have been instituted as a means of testing hypotheses about selective
deficits and, at the same time, ruling out generalized deficit explanations.
In a differential deficit design, the subject groups are given two closely related
tasks that differ on some hypothesized dimension. The results are evaluated by
comparing the groups in the pattern of performance over the tasks. Support for a
hypothesis is provided in finding a greater schizophrenic deficit on one task relative to
the other; in terms of an analysis of variance (ANOVA), this type of result
corresponds to a significant group by task interaction. Although the schizophrenic
group may show a deficit on both tasks, the observation of a differential impairment
across tasks suggests that performance is affected by factors other than a generalized deficit. Further, the differences between the tasks provide clues as to the cognitive processes in schizophrenia that deviate from normality. If, on the other hand, schizophrenic subjects show an equal deficit across tasks, their inferior performance is likely due to a generalized rather than to a schizophrenia-specific deficit. 12
In the Chapman (1958) card-sorting study, subjects were given problems con taining distractors of high associative strength as well as problems containing distrac- tors of low associative strength. These two types of problems, differing with respect to strength of associative distractors, correspond to the related tasks of a differential deficit design. Although the schizophrenic subjects showed a deficit in solving both types of problems, the deficit was significantly greater on problems involving strong associates relative to those involving weak associates—a finding consistent with
Chapman’s view that schizophrenia involves an excessive tendency to respond to associative distractors.
Chapman and Chapman (1973a, 1973b, 1978) later argued that a differential deficit design as described above can produce misleading results when the two tasks are not matched in terms of discriminating power. The discriminating power of a test refers to the ability of a test to truly differentiate more able from less able normal subjects. Two tests that have equal discriminating power are identical in terms of reliability and shape of the distribution of scores, as well as in terms of the mean, variance, and shape of the distribution of item difficulty. When tests that differ in discriminating power are given to subject groups that differ in overall ability, the group with a lower ability level may show a differential deficit across the tasks for two reasons. On the one hand, a differential deficit may be observed because the groups differ in terms of some specific ability that is differentially assessed by the two tasks.
Alternatively, a differential deficit may be observed because of a global deficit in the less able group. That is, the less able group may show a greater performance differ ence across the two tasks than the more able group solely because the tests differ with respect to their ability to differentiate less able from more able subjects. Hence, when tests that differ in discriminating power are employed in a differential deficit design, the main purpose for using the design—to yield information on specific versus global deficits—is defeated. For this reason, Chapman and Chapman (1973a, 1973b, 1978) have strongly advocated the use of differential deficit designs that include tasks matched on discriminating power.
In Chapman’s (1958) card-sorting study, normal subjects showed an increase in errors with increasing levels of associative strength, indicating that the problems that contained distractors of high associative strength were more difficult—and thus more discriminating—than the problems that contained distractors of low associative strength. Because the tasks were not matched on discriminating power, the finding of a differential deficit in the schizophrenic individuals may reflect a lower overall ability rather than a circumscribed deficit in a specific ability. With these issues in mind,
Rattan and Chapman (1973) devised a multiple-choice vocabulary test that included three types of items. In one type, one of the response alternatives included an associate of the vocabulary test word:
POOL means the same as: A. puddle (correct alternative) B. notebook (irrelevant alternative) C. swim (associative alternative) D. none of the above
Another type of item included no such associates among the response alternatives:
POINT means the same as: A. bum (irrelevant alternative) B. aim (correct alternative) C. sing (irrelevant alternative) D. none of the above
Rattan and Chapman (1973) pretested the two types of items using normal subjects— firemen and prison inmates—so that the discriminating power of the with-associates subtest matched that of the no-associates subtest. The investigators also devised a third type of item, for which the alternative “none of the above” was correct. The resulting vocabulary test, which included the three types of items arranged in a random order, was given to chronic unmedicated schizophrenic patients. 14
The results were analyzed in terms of the number of correct responses, expressed in ^-scores, on the with-associates and no-associates vocabulary items. Rattan and Chapman (1973) reported that the schizophrenic patients performed more poorly than normals on both types of items. More importantly, although the normal subjects did not differ in performance on the two types of items, schizophrenic individuals performed more poorly on the with-associates items than on the no associates items. Rattan and Chapman (1973) also selected a subgroup of normal subjects whose mean score on the no-associates vocabulary items was the same as that of the schizophrenic subjects. As was the case for the entire normal group, the normal subgroup showed the same level of performance on the with-associates and no associates items. Moreover, the schizophrenic subjects scored more poorly than the normal subgroup on the with-associates items. The use of tasks matched on discrim inating power rules out the possibility that the differential performance of the schizophrenic subjects was an artifactual result of the psychometric properties of the subtests. Thus, schizophrenic errors in the vocabulary task are attributable to an excessive tendency to select associatively related, although incorrect, responses. The results, then, are consistent with the theory of a schizophrenic accentuation of normal response biases.
Sentence Interpretation
Chapman, Chapman, and Miller (1964) studied schizophrenic performance in a verbal comprehension task involving double-meaning words. The word PLANT, for example, is a double-meaning word, or homograph. It most commonly refers to a living thing, and less commonly, to a factory. When homographs appear in discourse, individuals normally derive their intended meaning from the context in which they are used. Chapman et al. (1964) hypothesized that schizophrenic individuals, in encoun tering polysemous words, would be more likely than normals to interpret such words 15
with their most common meaning, irrespective of the context in which the words
occur. They presented chronic unmedicated schizophrenic patients and normal con
trols (firemen) with sentences containing homographs. Each sentence was followed by
a multiple choice test that assessed the subject’s interpretation of the homograph. For each homograph, two test sentences were constructed—one in which the homograph
was used in its most common sense, and the other in which the homograph was used
in its less common sense. Identification of the more and less commonly used meanings
of the homographs was predetermined through pilot testing with normals. An example
of a test item that employs the more preferred meaning of a double-meaning word is:
He couldn’t find bear meat because it was rare. This means: A. It was a meat with bones in it. B. It was an exceedingly uncommon meat. C. He likes partially cooked meat.
In this example, the word RARE is used in its most common form, that is, to designate
the uncommon quality of the meat. An example of a test item using the less preferred
meaning of a double-meaning word is as follows:
Robert said he likes rare meat. This means: A. He likes the kind of meat that is exceedingly uncommon. B. He likes a meat with bones in it. C. He likes partially cooked meat.
This example involves the use of the word RARE in the sense of being partly cooked.
For each type of item, the response options included a preferred-meaning, a nonpreferred-meaning, and an irrelevant response alternative. Subjects could thus respond to a test item with either the correct interpretation, a misinterpretation— choosing the preferred meaning response when the nonpreferred meaning was correct or vice versa, or an irrelevant answer.
The subjects’ performance was evaluated using a chance-corrected misinter pretation score, which was based on the number of raw misinterpretations minus the 16 number of irrelevant responses. In analyzing the corrected scores, Chapman et al.
(1964) reported that the schizophrenic subjects made more misinterpretations than normals on both types of sentences. The investigators examined performance on the two types of items in terms of a difference score, which was derived by subtracting the number of weak-meaning misinterpretations from the number of strong-meaning misinterpretations. Both subject groups showed a positive difference score, indicating that errors on items that employed the homographs in their less preferred sense outnumbered those on items that used the homographs in their more common sense.
However, the schizophrenic patients were found to have a significantly greater difference score than the normal controls.
The observation that schizophrenic patients made more errors than normals in responding to sentences involving the least preferred meanings is consistent with
Chapman et al.’s (1964) proposition that schizophrenic individuals are less sensitive to context than normals in interpreting words due to a bias towards making normatively high-frequency responses. However, the observation that the schizophrenic subjects made more errors than normals in responding to sentences involving preferred word meanings is not consistent with Chapman’s predictions. If, in theory, schizophrenic individuals err by choosing high-frequency alternatives, why would they misinterpret words used in the most preferred sense? Chapman et al. (1964) postulated that such results may be related to the fact that there is not always perfect agreement on the most preferred aspect of meaning of polysemous words. The results were thus reevaluated in light of the normative data used to determine the most and least preferred meanings of the double-meaning words. The homographs were assigned to one of two categories according to whether the normative data indicated a high or low degree of agreement as to the most preferred meaning. In examining performance on sentences that used the homographs in their most common sense, the schizophrenic and normal subjects 17
did not differ in terms of the number of misinterpretations in responding to high-
agreement items. Thus, schizophrenic individuals did not show a deficit in the
interpretation of homographs that were used in their most preferred sense, provided
that the homographs had a consensually dominant meaning. In contrast, on the low-
agreement items, the schizophrenic subjects committed a significantly greater number
of errors than the normals. Chapman and Chapman (1973a) also noted that the two types of test items may
have differed with respect to the ambiguity of the sentences. The test sentences had
been deliberately constructed so as to provide minimal cues regarding the proper
interpretation of the double-meaning words. Hence, the Chapman et al. (1964) results
may have been due to a greater degree of ambiguity in sentences that employed the
less preferred meanings relative to those involving the more preferred meanings of the
homographs. Chapman and Chapman (1973a) asked normal individuals to rate the test
sentences in terms of the ambiguity in meaning of the double-meaning words. They
found that the sentences using the less preferred meanings were in fact slightly but
significantly more ambiguous than sentences using the preferred meanings. The
original experiment included 19 of each type of sentence, and Chapman and Chapman
(1973a) eliminated 2 items of each sentence type so that the mean ambiguity ratings for the two types of sentences were equal. When the data analyses were repeated, identical results were obtained as in the original experiment Thus, performance of the
schizophrenic subjects in the verbal comprehension task did not appear to be due to differences in ambiguity between sentences using the more preferred and those using the less preferred meanings of words.
Another possible explanation for schizophrenic errors in the interpretation of double-meaning words is that schizophrenia may involve a loss of the less commonly used meanings of words. That is, the results of the Chapman et al. (1964) study may 18
have been mediated by a poorer knowledge of the meanings of words, rather than by a
predisposition for responding with preferred word meanings. In a separate experiment,
Chapman et al. (1964) presented chronic schizophrenic patients and normals with a
multiple-choice vocabulary test that included double-meaning words, single-meaning
words, and “filler” words. Each homograph was used to construct two test items: One
item assessed the subject’s knowledge for the word’s most preferred meaning and the
other item assessed knowledge for the less preferred meaning. The test items involving
single meaning and filler words (the response “none of the above” was correct for filler-word items) were included in order to make the purpose of the experiment less
apparent to subjects. The response alternatives of the test items did not include any
associative distractors, so that performance would not be affected by any bias towards
selecting associates. Chapman et al. (1964) found that the subject groups did not differ
in their ability to identify the less common meanings of the homographs. Thus,
schizophrenia does not appear to involve an actual loss of knowledge of the less
preferred aspects of word meanings. Hence, the results of the sentence interpretation
study suggest that when presented with homographs in context, schizophrenic
individuals appear to be less able than normal people to use contextual cues to guide
their interpretation of such words. Although they retain an adequate knowledge of the
less preferred meanings of double-meaning words, schizophrenic patients tend to
interpret such words in their most commonly used sense.
Summary The Chapmans have provided evidence for a schizophrenic accentuation of normal response biases in several experimental situations. In a concept classification task, schizophrenic patients show an excessive tendency to respond to verbal stimuli with associatively related but conceptually incorrect words. Schizophrenic errors on a vocabulary task can be similarly characterized, that is, as reflecting a bias towards 19
responding with associatively related words that are incorrect for the task at hand. In
interpreting homographs that appear in sentences, schizophrenic individuals show a
greater tendency than normals to disregard the contextual clues that ordinarily guide
interpretation, and instead, tend to respond with the most commonly used meanings of
words. In contrast, schizophrenic patients perform as well as normals in demonstrating
their knowledge of the less preferred meanings of homographs in vocabulary tasks that
do not include associative distractors among response alternatives.
Subsequent Work on the Chapman Theory
The Chapman studies have stimulated considerable interest in the idea that
cognitive difficulties in schizophrenia may be a result of an accentuation of normal
response biases. Over the years, a number of schizophrenia investigators have
conducted follow-up research in an effort to further evaluate the Chapman theory. The
bulk of these studies have involved the administration of either the tasks that Chapman
developed or close variations thereof. In contrast, relatively little work has been done
using procedures other than those devised by Chapman.
Follow-Up Studies Based on Chapman’s Tasks Card-Sorting Studies
Recall that the Chapman (1958) concept classification task involves presenting
subjects with sets of cards for sorting. Each set of cards includes a sorting-word card, containing a single word, and a guide-word card, containing three words. Subjects are
asked to indicate, for each set, which guide word belongs to the same conceptual class as the sorting word. For each item in the task, one of the guide words is an incorrect response alternative that is either a strong or weak associate of the sorting word.
Chapman and Chapman (1973a) cited two of Chapman’s (1958) results as support for the response bias theory. Schizophrenic subjects committed more errors relative to the 20
controls as the strength of the associative distractors increased. The same pattern of
results was obtained when the groups were compared in terms of a chance-corrected
associative error score, which consisted of the number of associative errors minus the
number of irrelevant response errors.
With a few exceptions, the card-sorting studies discussed below have evaluated results in a different manner from that employed by Chapman. Whereas Chapman
(1958) compared the performance of the subject groups on each of the two levels of
associative strength, subsequent investigators have generally compared group perfor
mance on the entire set of test items. This failure to consider performance on the two
levels of associative strength has not been accompanied by any clear rationale.
However, such a practice is likely to provide a conservative test of the Chapman
theory. Recall that the associative distractors in the low-strength condition were
essentially unrelated to their respective sorting words. A subject who has an excessive response bias thus would show a tendency to select associative distractors in the high-
strength condition, but not in the low-strength condition. Hence, the practice of combining data for the high- and low-strength conditions would be expected to dilute any effect attributable to an accentuated response bias and consequently, to decrease the probability that such a bias is detected.
Card-sorting performance in various subject groups. Chapman (1958) used his card-sorting task to examine associative distraction in chronic schizophrenic and medical patients. Replication studies involving other types of control groups have generally been interpreted as consistent with the notion that an accentuation of normal response biases is pathognomonic of schizophrenia. Marshall (1973) administered
Chapman’s card-sorting task to schizophrenic subjects, neurotic patients, and prison inmates. He reported that whereas the three subject groups did not differ in terms of the number of irrelevant errors, the schizophrenic individuals committed more 21 associative errors than both the neurotic patients and the prison inmates. Hemsley
(1976) also reported findings that were consistent with the Chapman theory. He found that acute schizophrenic subjects made more associative errors than depressed subjects on Chapman’s card-sorting task.
Feinberg and Mercer (1960), in contrast, argued that an excessive associative response bias may not be pathognomonic of schizophrenia. They administered
Chapman’s task to schizophrenic patients and to elderly individuals with organic brain disease. The two subject groups were compared with respect to the number of associa tive and irrelevant errors at each level of item difficulty and strength of associative distraction.2 Feinberg and Mercer (1960) reported that both subject groups exhibited an increase in errors with increasing levels of item difficulty and associative strength.
In each experimental condition, however, the patients with organic brain disease per formed more poorly than the schizophrenic subjects. Because the organic patients displayed the same pattern of errors as the schizophrenic subjects, Feinberg and
Mercer (1960) argued that an accentuation of normal response biases is present in disorders other than schizophrenia and hence, merely constitutes a nonspecific impairment.
Feinberg and Mercer’s (1960) conclusions seem less compelling, however, in light of the fact that the subject groups differed substantially in important ways.
Specifically, the average age of the organic patients was 69 years, whereas that of the schizophrenic subjects was 31. In addition, the subject groups, although matched on
IQ, almost certainly differed in terms of intellectual abilities. In order to obtain a given
IQ score, a younger person must exhibit a greater level of performance than an elderly person. The vast differences in age and intellectual abilities in the two subject groups may thus potentially explain the poorer card-sorting performance on the part of the organic brain syndrome group. This alternative explanation is consistent with the 22
observation that performance on the card-sorting task is significantly correlated with
intelligence (Hemsley, 1976; Hemsley & Hawks, 1974).
Dykes and McGhie (1976) utilized the card-sorting task in a study investi gating the relationship between schizophrenia and creativity. The investigators
administered the task along with other cognitive tasks to normal individuals who
evidenced either high or low degrees of creativity and to acute, nonparanoid
schizophrenic patients. The results were that both the schizophrenic and highly
creative individuals showed similar performance on tests involving divergent thinking. In contrast, on the card-sorting task, the schizophrenic group scored significantly
higher than both normal groups in terms of chance-corrected associative errors.
In sum, studies of card-sorting performance in various subject groups have
generally found that schizophrenic patients exhibit a greater tendency to respond to
associative distractors than do nonschizophrenic psychiatric patients, institutionalized
nonpsychiatric patients, and normal controls. These results suggest that an
accentuation of normal response biases is a unique characteristic of schizophrenia.
Card-sorting performance in various subtvpes of schizophrenic patients. Two studies have examined card-sorting performance in various subtypes of schizophrenia.
Gonen (1970) investigated the role of premorbidity and paranoid status in task perfor mance in a group of chronic schizophrenic patients. Although neither factor by itself was significantly related to associative error scores, the premorbidity by paranoid status interaction was significant. For good premorbid subjects, nonparanoid patients outperformed paranoid ones, whereas for poor premorbid subjects, paranoid patients outperformed nonparanoid patients.
A study by Hemsley and Hawks (1974) also deals with the subtype issue.
Hemsley and Hawks (1974) studied different combinations of schizophrenic subtypes than Gonen (1970) and hence, the findings of the two studies cannot easily be 23
integrated. Hemsley and Hawks (1974) found that on Chapman’s card-sorting task, acute and chronic schizophrenic patients did not differ in terms of associative errors
when differences due to age and estimated intelligence were covaried out.3 Subjects in
both groups committed very few irrelevant errors. Further, the correlation between
associative errors and a rating of paranoid symptomatology was not significant. This
result thus replicates Gonen’s finding that paranoid status was not in itself related to
performance. Three months after the initial testing, Hemsley and Hawks (1974)
retested the acute patients who remained hospitalized. The investigators reported that
the number of associative errors that were committed by this subgroup of patients did
not differ between the first and second testings.
Card-sorting tasks involving a gradation of levels of associative strength. Several studies have involved a modification of the Chapman (1958) task in terms of
the strength of associative distractors. Chapman included two levels of associative
strength in order to test the hypothesis that schizophrenic individuals show a greater response bias with increasing levels of associative strength. Roberts and Schuham
(1974) pointed out that Chapman’s low-strength distractors were in essence unrelated to their respective sorting words and argued that a more stringent test of the Chapman theory would involve the inclusion of a gradation of levels of associative distraction.
The investigators modified Chapman’s task by adding items with distractors of intermediate associative strength. They also studied the reliability with which normal raters ranked the associative strength of the distractors. The modified test contained only those items for which the reliability ratings were fairly high. Using this criterion, over half of the original Chapman items were eliminated. Roberts and Schuham
(1974) presented the modified task to chronic schizophrenic and alcoholic patients.
They found that on items containing distractors of low associative strength, the subject groups did not differ in terms of either the number of raw or chance-corrected 24 associative errors. In contrast, on items containing distractors of high associative strength, the schizophrenic subjects performed more poorly than the alcoholic subjects on both measures.
Paulman and Meyers-Abell (1980) reported comparable results in a study of paranoid schizophrenic patients, nonparanoid schizophrenic patients, and normal controls. Subjects were presented with the high- and moderate-associative distractor items developed by Roberts and Schuham (1974).4 Schizophrenic subjects obtained higher chance-corrected associative error scores in the high-distractor than in the moderate-distractor condition, whereas normal subjects did not display differential performance over the two distractor conditions. Paranoid and nonparanoid schizo phrenic subjects did not differ in their performance. Although Paulman and Meyers-
Abell (1980) did not examine the data in terms of an acute-chronic dichotomy, they reported that performance was not correlated with either cumulative hospital-stay time or number of admissions. These latter results suggest that chronicity does not mediate an excessive tendency to respond to associative distractors.
Speeded and slowed administration of the card-sorting task. The Chapman theory holds that schizophrenic and normal individuals commit the same types of errors, but to a different degree. In the card-sorting task, both schizophrenic and normal individuals commit errors that are predominantly associative in nature; however, schizophrenic individuals respond to incorrect associative distractors more than normal persons. Usdansky and Chapman (1960) were interested in identifying the conditions under which normal persons would show a heightened tendency to select associative distractors. The investigators postulated that normal individuals would show an increase in associative errors when completing the card-sorting task under time pressure than when completing the task under standard instructions. They divided the standard test items of the task into two sets and presented the sets to a group of 25
college students. For one set of items, subjects were asked to complete the items as
quickly as possible, while for the other set, subjects were instructed to complete each
item as carefully and accurately as possible. Usdansky and Chapman (1960) observed that an increase in time pressure was accompanied by an increase in associative and
irrelevant errors, as well as an increase in chance-corrected associative errors. Thus,
normal subjects exhibited a heightened normal response bias as a result of instructions
to speed their rate of responding to the card-sorting items.
Marshall (1973) hypothesized that schizophrenic individuals would also be
sensitive to instructions to speed as well as to slow their rate of responding to items in the card-sorting task. He predicted that in comparison to their performance under
standard, unpaced instructions, schizophrenic patients would exhibit an increase in
associative errors under time pressure and a decrease in errors under instructions to
slow their responses. In testing these predictions, Marshall divided the items of the
Chapman card-sorting test into 3 subtests and presented them to schizophrenic
patients. For one subtest, subjects were given the standard, unpaced instructions. For
the second subtest, subjects were asked to respond as quickly as possible. And for the
third subtest, subjects were asked to thoroughly and carefully think over each of their
responses. The unpaced and speeded subtests were also administered to neurotic patients and prison inmates. The results indicated that the subject groups did not differ
in terms of irrelevant errors in any of the instruction conditions. In the unpaced condition, schizophrenic patients committed more associative errors than both the neurotic patients and prison inmates, whereas the neurotic patients and inmates did not differ from one another. Schizophrenic subjects showed an increase in associative errors under the speeded instructions and a decrease in errors under the slowed instructions compared to their performance under standard instructions. As was the case in the Usdansky and Chapman (1960) study, the normal subjects showed an 26 increase in associative errors in the speeded condition compared to their performance in the unpaced condition. The neurotic patients exhibited a similar pattern of performance, with an increase in associative errors in the speeded relative to the unpaced condition.
Hemsley and Hawks (1974) administered the unpaced, speeded, and slowed versions of the card-sorting test to acute and chronic schizophrenic patients. The results were virtually identical to those of Marshall (1973). Subjects committed very few irrelevant errors under any of the instructional conditions. They showed an increase in associative errors under speeded instructions, and a decrease under slowed instructions relative to their performance under standard instructions.
Grimes and McGhie (1973) compared schizophrenic patients, psychiatric control patients, and normal individuals in the speeded version of the card-sorting task.
The experimental task varied from that of Marshall (1973) and Hemsley and Hawks
(1974) in that subjects were given time limits for completing the test items. Because subjects did not always complete the test items in the allowed time, the investigators examined errors in terms of the percentage of the total number of responses. In evaluating the results, the investigators reported only on differences that were significant at the .01 level.5 The three subject groups exhibited nearly identical performance in terms of the proportion of responses involving irrelevant errors, whereas the schizophrenic group scored twice as high as the other two groups in the proportion of associative responses. Although the schizophrenic group committed more associative errors than the control groups, the differences were not significant at a .01 level.
Studies that present speeded and slowed versions of the Chapman card-sorting task appear to report consistent findings. Schizophrenic patients, like normal individuals, show an increase in associative errors in response to an increase in time 27 pressure, and a decrease in such errors with a decrease in pressure. However, the examination of schizophrenic response biases under speeded and slowed versions of the card-sorting task does not appear to be particularly enlightening.6 Although the results of these studies suggest that the exaggerated response biases of schizophrenic patients are amenable to manipulation, the findings do not provide any clues as to how such biases are modified by task instructions.
Summary. In sum, follow-up studies involving the Chapman (1958) card- sorting task have generally produced evidence in accord with the hypothesis that schizophrenic patients show an accentuation of normal response biases. An exag geration of associative response biases in schizophrenic individuals has been docu mented in comparison to psychiatric control subjects (Hemsley, 1976; Marshall, 1973) as well as in comparison to various types of normal controls (Dykes & McGhie, 1976;
Hemsley, 1976). Further, this bias appears to characterize acute as well as chronic schizophrenic patients (Dykes & McGhie, 1976; Hemsley, 1976; Hemsley & Hawks,
1974; Roberts & Shuham, 1974). The influence of premorbidity and paranoid status on associative response biases is unclear, as few studies have been done in this area.
Schizophrenic individuals have been shown to exhibit a greater associative response bias with increasing levels of associative distraction (Paulman & Meyers-Abell, 1980;
Roberts & Shuham, 1974). Finally, an exaggeration in normal response biases is modifiable to some degree by experimental instructions to speed or slow the rate of responding (Grimes & McGhie, 1973; Hemsley & Hawks, 1974; Marshall, 1973).
Vocabulary Test Studies
Recall that Rattan and Chapman (1973) devised a multiple choice vocabulary test in which associative distractors served as response alternatives in some items, while no such associates were present in other items. The with-associates and no associates subtests were matched on discriminating power so that normal subjects 28
showed equal performance on the two subtests. Rattan and Chapman (1973) evaluated
the response bias theory by comparing the number of correct responses produced by
schizophrenic patients on the with-associates versus the no-associates subtests. They found that schizophrenic subjects performed more poorly on the with-associates than
the no-associates items, thus providing support for the theory.
Vocabulary test performance in various subject groups. Sengel and his
colleagues administered the Rattan-Chapman task to various subject groups in an
effort to study the interrelationship among diagnosis, thought disorder, and response biases (Sengel, Lovallo, Pishkin, Leber, & Shaffer, 1984). They argued that if, as
recent studies suggest, thought disorder is present in conditions other than schizo phrenia, then an excessive yielding to normal response biases should be related to the presence of thought disorder, regardless of diagnosis. Sengel et al. (1984) administered the vocabulary test to chronic paranoid schizophrenic patients, chronic nonparanoid schizophrenic patients, and manic patients. Subjects also received the Whitaker Index of Schizophrenic Thinking (WIST; Whitaker, 1980), which was used to measure thought disorder. The investigators found that on the WIST, the three groups did not differ in performance, suggesting that the schizophrenic and manic subjects evidenced similar degrees of thought disorder. On the Rattan-Chapman test, an analysis of the percentage of errors revealed no significant differences attributable to diagnostic group, subtest, or the interaction between group and subtest. WIST scores were significantly correlated with scores on the no-associates subtest as well as with scores on the combined subtests. The correlation between WIST performance and performance on the with-associates subtest was nearly significant.
Sengel et al. (1984) concluded that their results do not support the notion that an excessive associative response bias is pathognomonic of schizophrenia. Instead, they argued, an excessive bias is more strongly related to thought disorder than to 29 diagnosis. Although this study is laudable for its attempts to relate thought disorder to performance on an experimental task, the investigators overlooked a crucial finding in drawing their conclusions. Specifically, the finding of a nonsignificant subtest effect, along with a nonsignificant subtest by diagnosis interaction, indicates that the schizo phrenic subjects did not perform differentially on the with-associates and no-associates subtests. Thus, this study fails to replicate the basic findings of Rattan and Chapman
(1973). Sengel et al.’s (1984) very small sample sizes (n = 6 to 8) may have influenced the findings.
Braff and Saccuzzo (1985) compared schizophrenic and depressed patients on the Rattan-Chapman vocabulary test. The results indicated that although the schizo phrenic subjects made over twice as many associative errors as the depressed patients, the group differences were not significant. The authors did not examine performance on the with-associates versus no-associates subtests, thus limiting the conclusions that can be drawn. Thus, an associative bias may possibly characterize cognitive processes in depression as well as schizophrenia.
The next three studies examine vocabulary test performance in various subject groups in an attempt to evaluate the effect of institutionalization on normal response biases. Chapman studied schizophrenic patients who endured long periods of hospitalization. Individuals who are institutionalized on a long-term basis tend to live in highly regimented environments characterized by a dearth of stimulation. The question thus arises as to whether such a restricted and impoverished environment can in itself promote cognitive deficits. That is, the observation of a cognitive deficit in schizophrenic patients who experience lengthy hospitalizations may potentially be associated with schizophrenia, long-term institutionalization, or both. The next three studies take into account the potential effects of diagnosis and institutionalization by 30
including contrast groups composed of nonpsychiatric (medical) patients who are hospitalized for lengthy periods.
Klinka and Papageorgis (1976) administered the Rattan-Chapman vocabulary
test to schizophrenic patients, psychiatric patient controls, and medical patients.
Subjects were further identified as being hospitalized on either a short- or long-term
basis. The investigators reported that long-term patients of all diagnoses committed
more errors on the with-associates subtest than on the no-associates subtest. Short
term patients, on the other hand, did not show a differential pattern of performance
across the two subtests. No significant differences attributable to diagnostic group
were found. These results suggest that an accentuation of normal response biases may
be observed in a variety of individuals who are hospitalized for extended periods of
time. Klinka and Papageorgis (1976) also examined their data informally for ways in
which the performance of schizophrenic patients might be distinguished from that of
patients with other diagnoses. In considering the proportion of subjects in each group
who exhibited a pronounced response bias, the long-term schizophrenic patients stood
out. Half of the long-term schizophrenic subjects showed such a bias, whereas less
than 30 percent of subjects from the other groups did so. In addition, Klinka and
Papageorgis (1976) noted that if they had analyzed their data using l-tests—as Rattan and Chapman (1973) did—instead of ANOVAs, the outcome would have indicated that the short-term schizophrenic group, as well as all of the long-term groups, committed more errors on the with-associates subtest than on the no-associates subtest.
The formal analyses, however, indicate that an excessive response bias is associated with long-term institutionalization and chronic illness rather than with diagnosis.
Klinka and Papageorgis (1984) were interested in studying the effect of length of institutionalization on vocabulary test performance in paranoid and nonparanoid schizophrenic patients and in medical patients. All of the patient groups had chronic 31
(i.e. long-term) illnesses, but differed in terms of length of hospitalization. Half of the
patients in each diagnostic group were inpatients who had been hospitalized for many years. The other half were outpatients who had been previously hospitalized for only
brief periods. Instead of analyzing results by comparing performance on the with-
associates versus no-associates subtests, the investigators examined performance on
the with-associates subtest with performance on the no-associates subtest serving as a
covariate. Klinka and Papageorgis (1984) reported that inpatients performed more
poorly than outpatients, and that schizophrenic patients performed more poorly than
medical patients. Of the schizophrenic patients, nonparanoid subjects performed more
poorly than paranoid ones. However, the diagnosis by hospitalization-status inter
action was not significant. Because the nature of the data analyses preclude conclu
sions as to whether a particular subject group did or did not exhibit an accentuation of
normal response biases, confident assertions can only be made as to whether one
subject group exhibited more of a bias than another. Klinka and Papageorgis’ (1984) findings thus suggest that long-term institutionalization is associated with a greater response bias than short-term institutionalization, that schizophrenia is associated with a greater bias than nonpsychiatric medical illness, and that nonparanoid schizophrenia is associated with a greater bias than paranoid schizophrenia.
Saccuzzo and Braff (1980) presented the Rattan-Chapman test to five groups of subjects: chronic schizophrenic patients, institutionalized elderly patients, noninstitu tionalized elderly individuals, psychiatric patient controls, and normal controls. Both schizophrenic and institutionalized elderly subjects produced more errors in respond ing to the with-associates items than to the no-associates items, whereas the other groups showed no differential performance across subtests. The institutionalized elderly subjects were individuals who required nursing-home care due to an inability to function independently, but were without obvious cognitive deterioration. Based on 32
these results, Saccuzzo and Braff (1980) suggested that schizophrenic and elderly
individuals may share a common cognitive deficit. They speculated that the aging
process, combined with institutionalization, may be related to a disturbance in
associative functioning. However, they noted, the cognitive processes that underlie an
associative response bias in schizophrenic patients and the institutionalized elderly
may or may not involve similar mechanisms.
Studies that have attempted to evaluate the effects of institutionalization on
vocabulary task performance have thus consistently reported that long-term institu tionalization is associated with an excessive yielding to normal response biases.
Callahan and Saccuzzo (1986) studied the performance of first-degree relatives
of schizophrenic patients on the Rattan-Chapman vocabulary task. The study of
biological relatives of schizophrenic individuals is an increasingly used technique that
has its basis in evidence for a genetic contribution to schizophrenia (Gottesman &
Shields, 1972; Heston, 1966; Rosenthal, 1970). The technique is utilized to identify
“markers” of schizophrenia, that is, features that accompany the genetic aspects of the
disorder (Harvey, Walker, & Wielgus, 1986; Zubin & Spring, 1977). In addition, the
technique has the advantage of ruling out the influence of such factors as repeated
hospitalization and long-term medication usage, which plague interpretations of direct
observations of schizophrenic subjects. On the Rattan-Chapman vocabulary test,
Callahan and Saccuzzo (1986) found that first-degree relatives of schizophrenic
persons performed more poorly than demographically matched controls on the with-
associates subtest, whereas the two groups did not differ on the no-associates subtest.
Moreover, in the blood relatives group, the difference between with-associates and no
associates performance was larger than that of Rattan and Chapman’s (1973) schizo phrenic subjects. According to Callahan and Saccuzzo (1986), the finding that
schizophrenic individuals and their biological relatives showed similar deficits on the 33 vocabulary task provides support for the notion that the Chapman theory deals with a fundamental aspect of schizophrenia rather than with the effects of institutionalization, medication, and the like. Callahan and Saccuzzo (1986) further suggested that an associative response bias may serve as a cognitive marker that is present in individuals who are genetically predisposed to schizophrenia.
Vocabulary test performance in various subtypes of schizophrenic patients. No studies have examined schizophrenic performance on the Rattan-Chapman task in regard to measures of chronicity or premorbidity. Two studies, both of which were introduced above, have compared paranoid with nonparanoid schizophrenic patients on the vocabulary task. Sengel et al. (1984) found that paranoid and nonparanoid schizophrenic patients did not differ in performance. However, the investigators did not observe an increase in errors on with-associates compared to no-associates items, and hence, failed to document an exaggeration in response biases in either schizo phrenic group. Klinka and Papageorgis (1984) found that although both nonparanoid and paranoid schizophrenic patients performed more poorly on the with-associates than on the no-associates subtest, the nonparanoid group showed a greater differential performance across subtests than the paranoid group.
Variations of the vocabulary task. Boland and Chapman (1971) evaluated the
Chapman theory using a vocabulary test that is similar to the one developed by Rattan and Chapman (1973). Like the Rattan-Chapman test, the Boland-Chapman test is made up of items presented in a multiple-choice format, with some items containing associative distractors as response alternatives. In contrast to Rattan and Chapman
(1973), Boland and Chapman (1971) did not match the with-associates and no associates subtests on discriminating power. In fact, they viewed the no-associates items as “filler” items that were included in order to make the purpose of the 34
experiment less apparent to subjects. As such, performance on the no-associates portion of the vocabulary test was not examined in the data analyses. Instead, Boland
and Chapman (1971) compared, within each subject group, the number of correct
responses relative to the number of associative distractor responses. The investigators
found that normal subjects chose more correct responses than associative distractors,
whereas chronic unmedicated schizophrenic subjects chose more associative distrac
tors than correct responses. Thus, schizophrenic patients showed a preference for
endorsing associatively related but incorrect responses. The investigators concluded that the schizophrenic subjects’ tendency to endorse associatively related but incorrect
responses suggests that, in accord with the Chapman theory, schizophrenia involves an
exaggeration of normal response biases and a failure to consider context.
Hamsher and Arnold (1976) administered the Boland-Chapman vocabulary test
to schizophrenic patients and psychiatric patient controls. The two subject groups did
not differ in terms of number of correct responses, number of incorrect associative
distractor responses, or number of irrelevant responses. Both of the subject groups
performed somewhat better than Boland and Chapman’s schizophrenic patients, but
worse than their normal controls. Hamsher and Arnold (1976) concluded that perfor
mance on the Boland-Chapman test is not specifically related to thought disorder or
schizophrenia. The omission of a normal control group, however, does not allow the
investigators to rule out the possibility that the performance of their patient groups
may not have differed from that of normal individuals. Such a finding would constitute
a failure to observe an accentuation of normal response biases in the patient groups. In addition, Hamsher and Arnold’s (1976) schizophrenic subjects appear to be of the acute subtype, as their cumulative hospitalization averaged less than six months.
Hence, the no-difference results might suggest that acute schizophrenic patients do not 35 differ from nonschizophrenic psychiatric patients in the tendency to endorse associatively related but incorrect responses to vocabulary test items.
Summary. Attempts to replicate and extend Rattan and Chapman’s vocabulary test study have produced several interesting findings. For one, an associative response bias has been documented in certain select groups of control subjects as well as in schizophrenic subjects. The results of Klinka and Papageorgis (1976), Saccuzzo and
Braff (1980), and Klinka and Papageorgis (1984) imply that an accentuation of normal response biases may be characteristic of individuals—including medical patients— who endure long periods of institutionalization. However, this conclusion needs to be qualified by other findings indicating that schizophrenia per se is also linked to an associative response bias. The Klinka studies indicated that although long-term inpatients exhibit an associative response bias, schizophrenic patients hospitalized on a short-term basis may also show such a bias. Further, the chronic schizophrenic patients in Klinka and Papageorgis’s (1976) study were distinguished from other long-term patient groups in showing an excessively pronounced bias.
The vocabulary test studies have also raised the issue of whether an associative response bias is observed in psychiatric conditions other than schizophrenia. Using a mixed-diagnosis psychiatric control group, Hamsher and Arnold (1976) found no difference between schizophrenic patients and controls. The investigators do not report whether both groups failed to show an associative response bias, which would consti tute a failure to replicate Chapman’s findings of a bias in schizophrenic patients, or whether both groups did show a bias, which would suggest that the bias phenomenon is not pathognomonic of schizophrenia. On the other hand, Saccuzzo and Braff (1980) found that schizophrenic patients showed more of a normal response bias in comparison to a mixed-diagnosis psychiatric control group. In studies involving homogeneous psychiatric controls, both Sengel et al. (1984) and Braff and Saccuzzo 36
(1985) reported no-difference findings when comparing schizophrenic patients with, respectively, manic and depressed patients. Both of these studies were noted to have problems, however. Sengel et al. (1984) employed very small samples and failed to document an associative bias in even the schizophrenic patients. Braff and Saccuzzo
(1985) examined absolute level of associative errors without reference to errors on the no-associates subtest.
Callahan and Saccuzzo’s (1986) finding of an associative response bias in first- degree relatives of schizophrenic individuals is particularly intriguing. The presence of schizophrenic-like behavior in nondisturbed, never-hospitalized, blood relatives of schizophrenic patients is compelling evidence for the notion that an accentuation of normal response biases is associated with schizophrenia and not merely an epiphe- nomenon due to institutionalization.
Sentence Interpretation Studies
Recall that Chapman, Chapman, and Miller (1964) examined the performance of chronic schizophrenic patients and normal individuals on a sentence interpretation task. Subjects were presented with test sentences, each of which contained a multi meaning word. Some sentences utilized the homographs in their most commonly used
(strong) sense, and others utilized the less-preferred (weak) meanings of the homo graphs. Following each sentence, subjects completed a multiple-choice test that assessed their interpretation of the sentence homograph. Subjects’ responses were scored in terms of the number of strong- and weak-meaning misinterpretations, that is, the number of strong-meaning responses on items in which the weak meaning was correct and vice versa. These misinterpretation scores were then subjected to a correction for guessing by subtracting the number of irrelevant responses from the number of misinterpretations. Chapman et al. (1964) compared the subjects groups on a score of the difference between the chance-corrected strong and weak 37
misinterpretation scores. They found that schizophrenic subjects had higher difference
scores than normal subjects. Thus, schizophrenic subjects exhibited a greater tendency
than normals to misinterpret homographs in context by responding with the most
commonly used meanings of words.
Sentence interpretation performance in various subject groups. Neuringer, Fiske, Schmidt, and Goldstein (1972) compared chronic schizophrenic patients with
psychiatric patient controls on Chapman et al.’s (1964) sentence interpretation task.
The groups did not differ in terms of the number of strong or weak misinterpretations.
Moreover, for both subject groups, the number of strong misinterpretations and the
number of weak misinterpretations did not differ. Hence, contrary to the Chapman
theory, neither subject group in Neuringer et al.’s (1972) study showed a heightened tendency to interpret words in their most commonly used sense. A study by Blaney
(1974) also failed to replicate Chapman et al.’s (1964) findings. Blaney presented the sentence interpretation task to schizophrenic subjects, psychiatric control patients, and hospital staff. The results were that neither patient group differed from the normal group in terms of their chance-corrected difference scores. In contrast, Williams,
Hemsley, and Denning-Duke (1976) reported findings that were consistent with those of Chapman et al. (1964). They found that schizophrenic patients obtained higher chance-corrected difference scores on the sentence interpretation task than did hospital staff, who served as normal controls.
Persons and Baron (1985) used the sentence interpretation task to evaluate the extent to which Chapman’s theory accounts for thought disorder in psychiatric patients in general. They administered the task to psychiatric patients who were categorized as thought-disordered or non-thought-disordered according to the presence of incoher ence in their speech. Incoherent speech is generally regarded as a severe symptom of thought disorder, and has been defined as speech that is essentially incomprehensible 38 due to the presence of gross grammatical irregularities, tenuously connected or com pletely unrelated ideas, numerous irrelevant ideas, and words whose meanings are used in a bizarre fashion (American Psychiatric Association, 1987). Approximately 75% of Person and Baron’s (1985) total subject sample was constituted by patients with a schizophrenia-related diagnosis, and approximately 20% of the patients were assigned a diagnosis of major depression. The experimenters found that the two groups did not differ in terms of strong misinterpretation errors, weak misinterpretation errors, or the difference between strong and weak misinterpretations. Hence, the presence of one form of thought disorder—namely, incoherence—was not reliably associated with an exaggerated normal response bias.
Sentence interpretation performance in various subtvpes of schizophrenic patients. Blaney (1974) reported that schizophrenic patients who had a history of lengthy hospitalization obtained higher chance-corrected difference scores on the sentence interpretation task than schizophrenic subjects with a relatively short hospitalization history. He also found that schizophrenic individuals who were rated as high on a scale of florid psychosis obtained higher chance-corrected difference scores than those who received low ratings. Williams, Hemsley, and Denning-Duke (1976) also compared acute and chronic schizophrenic patients on the sentence interpretation task and obtained results that were virtually identical to those of Blaney.
Studies that examine the relationship between performance on Chapman et al.’s
(1964) sentence interpretation task and premorbidity and paranoid status are not available.
Other sentence interpretation tasks involving homographs. Several investi gators have devised sentence interpretation tasks that are patterned after Chapman et al.’s (1964) task. Benjamin and Watt (1969) were interested in comparing the response 39
bias theory with the theory that schizophrenic individuals have difficulty in abstract
thinking (Bolles & Goldstein, 1938; Vigotsky, 1934). Benjamin and Watt’s (1969) task contained sentences that varied not only in terms of whether the correct interpre tation of the homographs involved a strong or weak meaning, but also in terms of
whether the correct interpretation of the homographs involved an abstract or concrete
meaning. The investigators presented the task to chronic schizophrenic patients,
patients with organic brain syndrome, alcoholic patients, and normal controls. All
subject groups showed a tendency to select response alternatives that involved a
concrete interpretation of the homographs. This concreteness bias was not differen
tially present in the four groups, suggesting that schizophrenic persons are not excessively prone to interpreting words in a concrete fashion. The analyses also revealed that none of the groups showed a primacy (strong-meaning) bias, that is, a
greater tendency to commit strong-meaning than weak-meaning errors. However, when the test items were dichotomized according to their degree of ambiguity, and the analyses were repeated, some interesting differences emerged. On high-ambiguity sentences, subjects tended to commit errors by choosing weak-meaning responses, and this weak response bias did not differ among the groups. On low-ambiguity items, subjects of all groups showed a strong response bias; however, the schizophrenic group exhibited a greater primacy bias than the other groups. Benjamin and Watt
(1969) interpreted these results as consistent with the response bias theory. The investigators speculated that in responding to highly ambiguous sentences, subjects may have adopted a strategy in which they assumed that a sentence with a vague meaning contains words that are used in an unusual manner. This response strategy thus resulted in a tendency to endorse weak-meaning responses on ambiguous items.
In contrast, in responding to relatively unambiguous items, subjects committed errors primarily by endorsing strong-meaning alternatives. 40
Miller (1974) also constructed a sentence interpretation task that, like Benjamin
and Watt’s (1969) task, involved two levels of concreteness combined with two levels of meaning strength. The task was administered to chronic and acute schizophrenic
patients and to normal subjects. The three groups did not differ in the number of
irrelevant errors. In contrast to Benjamin and Watt’s (1969) subjects, who exhibited a
tendency to endorse concrete meaning responses, Miller’s subjects showed a tendency
to select abstract meaning responses. The subject groups did not differ in their
preference for selecting abstract over concrete responses and thus, the results were
inconsistent with the notion that schizophrenic thought is overly concrete. Miller also
found that the subject groups did not differ in terms of the number of strong-meaning
or weak-meaning errors. All groups showed a primacy bias in that they committed
more strong-meaning than weak-meaning errors. However, this primacy bias did not
significantly differ among the groups. Like Benjamin and Watt (1969), Miller
examined the data in terms of the ambiguity of the test sentences. On high-ambiguity
items, the three subject groups did not differ with respect to the total number of errors.
In explaining these results, Miller speculated that high-ambiguity items may have been
so difficult that they did not differentiate between the subject groups. In contrast, on low-ambiguity items, the schizophrenic groups committed significantly more overall errors than the normal control group. Miller did not conduct analyses to determine whether this elevated error rate in the schizophrenic groups was attributable to a tendency to commit strong-meaning errors. Hence, no definitive conclusions can be drawn regarding the group differences in responding to the low-ambiguity items.
Neuringer, Kaplan, and Goldstein (1974) constructed a sentence interpretation task that employed emotionally charged and neutral homographs. They found that both chronic schizophrenic patients and psychiatric patient controls committed more weak errors on emotionally charged items than on neutral items. More importantly, the 41
schizophrenic group committed more weak errors than the control patients, whereas the two groups did not differ in the production of strong errors. Thus, a strong response bias was not observed in either patient group. In fact, the finding of a weak-meaning
bias in the schizophrenic patients is diametrically opposite to what would be expected
from the Chapman theory.
Strauss (1975) utilized a sentence interpretation task to examine the effects of
chronicity and medication on response biases. He hypothesized that a strong-meaning
response bias would be more exaggerated in unmedicated than in medicated
schizophrenic patients, and in chronic relative to acute patients. He administered the
task, composed of items from Chapman et al.’s (1964) and Benjamin and Watt’s
(1969) procedure, to four groups of subjects: unmedicated chronic schizophrenic
patients, medicated chronic schizophrenic patients, medicated acute schizophrenic
patients, and prisoners. All groups exhibited a strong meaning response bias, as
evidenced by a greater number of strong-meaning than weak-meaning errors. The
groups did not differ in terms of the number of weak or irrelevant errors. However,
both of the chronic schizophrenic groups committed more strong errors than the other
groups, suggesting that the strong-meaning response bias of the chronic groups was more extreme than that of the other groups. These results suggest that an accentuation of normal response biases is found in chronic but not acute schizophrenic patients, and that medication does not ameliorate this cognitive aberration.
Naficy and Willerman (1980) constructed a sentence interpretation task that included sentences with single-meaning as well as double-meaning words. The task was administered to paranoid schizophrenic patients, manic patients, and mental health workers. On sentences that utilized the preferred meanings of the homographs as well as on sentences that contained single-meaning words, the subject groups did not differ in the number of correct responses, the number of misinterpretations, or the number of 42
irrelevant responses. On sentences that used the homographs in their least preferred
sense, schizophrenic and manic patients made more misinterpretations than did normal
controls, whereas the three groups did not differ in the number of irrelevant responses.
Naficy and Willerman (1980) concluded that an excessive yielding to normal response
biases is a feature of manic disorder as well as of schizophrenia.
Taken together, studies that have utilized a sentence interpretation task other
than the one developed by Chapman et al. (1964) have reported inconsistent findings.
Two of the studies reviewed have produced evidence that schizophrenic patients are
especially prone to interpret homographs in their most commonly used sense (Naficy
& Willerman, 1980; Strauss, 1975). Another study documented an excessive strong-
meaning bias in schizophrenic patients, but only after the ambiguity of the test items
was taken into account (Benjamin & Watt, 1969). Yet another study found differences between schizophrenic patients and controls after controlling for ambiguity, but data analyses did not allow conclusions to be drawn as to whether the schizophrenic-control differences involved a predisposition for selecting strong-meaning responses (Miller,
1974). Finally, one study did not consider the ambiguity of test items in evaluating performance, and failed to document a strong-meaning bias in any subject group
(Neuringer, Kaplan, & Goldstein, 1974). Recall that Chapman and Chapman (1973a) found that the ambiguity of test items in a sentence interpretation task affected subjects’ responses. Such a finding is of little surprise, as the ambiguity of a test item essentially refers to its difficulty level. The conflicting results of replication studies that utilized various sentence interpretation tasks thus may have been due to differ ences in ambiguity between test items in which the homographs were used in their most common sense and those in which the homographs were used in their less- preferred sense. In other words, the discriminating power of the strong-meaning and weak-meaning sentences may have varied within and between studies, resulting in 43 inconsistent findings across studies. Another possible reason for the differential results
has to do with the fact that some studies manipulated the concreteness or the affectivity of the intended meaning of the homographs. Subjects may not show the
same kind of response biases in interpreting test items that closely resemble those of
Chapman et al.’s (1964) task as they do in interpreting items that vary in concreteness
and affectivity.
Schizophrenic patients’ knowledge of secondary word meanings. Several
studies have confirmed Chapman et al.’s (1964) assertion that a primacy bias in
schizophrenic persons’ performance on sentence interpretation tasks is not attributable
to a lack of knowledge of less common word meanings. Miller (1974) found that both
acute and chronic schizophrenic subjects showed normal levels of performance on a
multiple-choice vocabulary test that assessed knowledge of the secondary meanings of
homographs. Deckner, Greenberg, and Cash (1971) presented schizophrenic and medi
cal patients with a homograph vocabulary test in a free-response format. Subjects were
presented with homographs and cue words, which were either strongly or weakly
associated with the secondary meanings of the ambiguous words. They were told that
the to-be-defined words were words with more than one meaning and were asked to
use the cue words to determine which meaning to give in their responses. For each
item, the cue word was presented first followed by the homograph, which was given
either 1 second or 30 seconds after the presentation of the cue. The delayed-
presentation condition was designed to examine the possibility that schizophrenic
individuals might drift towards a primacy bias when required to retain a response for
an otherwise unoccupied interval. The subject groups did not differ in the extent to
which they gave primary and secondary meanings in their definitions. Subjects in both
groups made more errors when the cues were weak than when they were strong, and the effect of the two types of cues did not differ in the two groups. Further, the 44
immediate versus delayed condition did not result in differential performance across
groups. Thus, schizophrenic subjects showed an adequate knowledge of the less com
monly used aspects of word meanings. Further, when specifically instructed as to the
multiple-meaning nature of homographs, they were able to use weak cues to prompt retrieval of secondary meanings. These results suggest that schizophrenic individuals
can use context in interpreting words when instructions encourage them to do so.
Schizophrenic patients’ interpretation of syntactically ambiguous material.
Blaney (1974) was interested in whether schizophrenic individuals exhibit a primacy bias in processing phrases that are ambiguous in their grammatical structure, and that
depend on context for their proper interpretation. Blaney provides the following
example:
THE PACKAGE WAS LEFT BY THE MAN IN THE GREEN SWEATER.
This example might mean that a man wearing a green sweater left a parcel somewhere.
Alternatively, it might indicate that a package was placed beside a man wearing a green sweater. With no other context, normal individuals tend to respond to the sentence with the first interpretation. Now consider the following sentences:
THE PACKAGE WAS LEFT BY THE MAN IN THE GREEN SWEATER, NOT BY THE DELIVERY BOY.
THE PACKAGE WAS LEFT BY THE MAN IN THE GREEN SWEATER, NOT BY THE TREES.
In the first sentence, the intended meaning of the phrase THE PACKAGE
WAS LEFT BY THE MAN IN THE GREEN SWEATER is the same meaning that normals prefer when the phrase is encountered out of context In the second sentence, the less preferred meaning is the intended interpretation. Blaney (1974) constructed a sentence interpretation task similar to Chapman et al.’s (1964), except that the sen tences contained syntactically ambiguous phrases rather than semantically ambiguous homographs. As in Chapman et al.’s (1964) task, a multiple-choice test followed each 45
sentence. The response selections of each multiple-choice test included a strong- meaning alternative, a weak-meaning alternative, and an irrelevant alternative. And
like Chapman et al. (1964), Blaney evaluated his results in terms of a chance-corrected
difference score, which was based on the number of strong misinterpretations minus
the number of weak misinterpretations. He found that schizophrenic patients did not
differ from either psychiatric patients or normal controls in their performance on the
task. These results suggest that the Chapman hypothesis does not extend to
syntactically ambiguous material. Schizophrenic patients appear to adequately use
context in comprehending phrases that have multiple meanings.
Summary. Although a number of the sentence interpretation studies have
conformed to the Chapman theory, a number of others have not. In comparison to the
card-sorting and vocabulary test studies, the sentence interpretation studies have fared
the least well in providing support for the response bias theory. As the studies differ
along a number of dimensions (e.g., type of control groups studied, demographic variables used to produce matched subject groups, criteria for establishing diagnosis), the reasons for the variable results is not readily apparent. The replication failures do not seem to be a product of the specific sentence interpretation task administered, as variable results have been reported across studies that used Chapman et al.’s (1964) task as well as across those that used other sentence interpretation tasks. One possi bility is that a heightened bias toward interpreting homographs in their most common sense may not characterize all schizophrenic individuals. Or perhaps, as will be discussed below, the presence of thought disorder rather than of schizophrenia per se is associated with an accentuated response bias. However, the presence of one type of thought disorder symptom—incoherence—has not been shown to distinguish patients with an exaggerated normal response bias (Persons & Baron, 1985). 46
Several follow-up studies have suggested that chronicity may mediate a response bias in the case of interpreting double-meaning words in context. An
excessive primacy bias in chronic but not acute schizophrenic patients has been reported in studies using Chapman et al.’s (1964) test (Blaney, 1974; Williams,
Hemsley, & Denning-Duke, 1976) as well as in studies using other sentence interpre
tation items (Strauss, 1975). Miller (1974), in contrast, using his own sentence
interpretation task, reported that acute and chronic schizophrenic groups did not differ
from each other. However, Miller also found that neither of the schizophrenic groups
differed from a normal control group.
The sentence interpretation studies have produced contradictory conclusions
as to whether a primacy bias is characteristic of psychiatric patients other than those
with schizophrenia. Some studies have documented an accentuated bias in chronic
schizophrenic patients along with a lack of a bias in psychiatric patient controls
(Benjamin & Watt, 1969; Blaney, 1974). Others have reported that schizophrenic
subjects and psychiatric patient controls do not differ in terms of primacy response
tendencies (Naficy & Willerman, 1980; Neuringer, Fiske, Schmidt, & Goldstein,
1972; Neuringer, Kaplan, & Goldstein, 1974). One clue as to the possible reason for
these contradictory results is given by Naficy and Willerman (1980), who employed a patient control group composed of individuals with a diagnosis of mania. With the exception of Benjamin and Watt (1969), the other studies utilized heterogeneous control groups, composed of nonschizophrenic psychiatric patients with a variety of diagnoses. Naficy and Willerman (1980) found that both schizophrenic and manic patients showed a primacy bias, suggesting that similar cognitive deficits may be present in schizophrenia and mania. They argued that these results are consistent with evidence, given in descriptive studies of thought disorder symptoms
(Andreasen, 1979b), that manic patients, like schizophrenic patients, may exhibit 47 signs of thought disorder. The presence of thought disorder, rather than
schizophrenia, may thus be associated with an excessive primacy bias in the interpretation of words.
Other Lines of Evidence
The present section will focus on studies that have evaluated the response bias theory using empirical strategies other than those developed by Chapman.
Recognition Memory
Mourer (1973) predicted that a bias toward responding to the strong meaning aspects of words would result in certain types of errors in a recognition memory task.
He reasoned that if schizophrenic patients respond to words primarily in terms of strong meanings, then their errors in a recognition test ought to be characterized by a tendency to falsely recognize words that, although never presented, share a strong- meaning response with the to-be-remembered material. The words PIG and DOG, for example, share a strong-meaning response in that normal persons tend to include the statement IS AN ANIMAL in their definitions of both words. Further, normal indi viduals tend to rank the statement IS AN ANIMAL as the most important statement in defining both words. In contrast, the words NEWSPAPER and MAGAZINE share a relatively weak-meaning response. Although normal persons list the statements
READING MATERIAL and INFORMATIVE in defining both words, they do not consistently rank either of these shared meaning statements as first in importance.
Mourer predicted that if schizophrenic patients tend to respond to words primarily in terms of strong-meaning responses, they should be particularly prone to confuse words that share the same strong meaning.
In testing his predictions, Mourer (1973) selected a set of word pairs in which the members of each pair shared a strong- or weak-meaning response. One member of 48
each pair was chosen to serve on a list of to-be-remembered words. Subjects were
presented with the list of words and then asked to complete a recognition test. The test
consisted of a random arrangement of all of the words from the word-pair set inter
mixed with control words that were unrelated to the word pairs. Mourer found that
acute schizophrenic patients falsely recognized more of the strong- than weak-meaning
distractors, whereas normals did not exhibit a differential error rate for the two types of
distractors. These results suggest that schizophrenic individuals appear to retain
information in memory primarily in terms of the strong-meaning aspects of words.
Although normal individuals are also likely to retain this type of information, they
presumably rely on other types more than do schizophrenic individuals.
Recall Memory
Watson and Plemel (1981) argued that an accentuation of normal response
biases would result in a tendency to confuse to-be-remembered material with
associatively related information that is present at the time of study. They presented
schizophrenic patients, psychiatric patient controls, and normal individuals with three
recall tasks. The three tasks each involved the presentation of a list of 150 to-be- remembered words. In two of the tasks, distractor words were presented orally while
the to-be-remembered words were presented visually. In one of the distractor con ditions, the distractors were unrelated to the to-be-remembered words. In the other distractor condition, words that were associates of the list words served as distractors.
The three tasks were matched on discriminating power in order to render performance measures that would be comparable across the tasks. Watson and Plemel (1981) reported that distraction resulted in a greater detriment in performance in the
schizophrenic group relative to the other groups, but that associative and irrelevant distractors did not affect performance differentially in schizophrenic subjects. Further, schizophrenic performance was not significantly related to premorbidity, paranoid 49 status, or length of hospitalization. The investigators concluded that the results are more consistent with an interference theory of cognitive dysfunction in schizophrenia rather than an associative bias theory. That is, they suggested that although their results do not necessarily discount the Chapman theory, they do indicate that the theory may not account for some cognitive deficits in schizophrenia.
Pronunciation of Nonhomophonic Homographs
Persons and Baron (1985) attempted to extend Chapman’s work on homograph interpretation to the study of homograph pronunciation. They focused on homographs that are not homophonic, that is, words that have more than one pronunciation (e.g.
LEAD, REFUSE). Persons and Baron (1985) cited evidence that normal individuals show a longer latency to respond in pronouncing such words than in pronouncing control words. Presumably, normal people consider the various possible pronun ciations of ambiguous words, resulting in longer response times, or an ambiguity effect. Persons and Baron (1985) hypothesized that if thought disorder is due to an exaggerated normal response bias, it should be associated with a tendency to rely on the most commonly used pronunciations of nonhomophonic homographs. That is, thought-disordered individuals ought to show less of an ambiguity effect than normals in responding to phonetically ambiguous words. In testing this hypothesis, Persons and
Baron (1985) constructed two types of nonhomophonic homograph lists. In one type, cue words appeared before homographs. The cue words were selected so as to increase the probability that the less common pronunciations of the homographs would be made
(e.g. GARBAGE before REFUSE). The other type of homograph list contained no such cues. These homograph lists were presented to subjects along with control lists, consisting of nonhomographic words, and subjects were asked to read each list as rapidly as possible. The subjects were psychiatric patients with a variety of diagnoses, grouped as either thought-disordered or non-thought-disordered on the basis of 50
incoherence ratings. Persons and Baron (1985) measured the time taken to respond to
the homograph and control lists, and took the difference between the two as a measure of ambiguity. The results were that for all groups, subjects took longer to read the
homograph lists relative to the control lists, and the groups did not differ in terms of an
ambiguity effect. Thus, an accentuation of normal response biases does not appear to
mediate the pronunciation of phonetically ambiguous words in thought-disordered—or
more precisely, incoherent—psychiatric patients.
Semantic Priming
Swinney and his associates have conducted a series of studies that examine the
processes by which normal individuals access meaning from words (e.g. Swinney,
1979, 1984; Swinney, Onifer, Prather, & Hirshkowitz, 1979). In one experiment,
Swinney (1984) studied the manner in which context guides lexical access in
schizophrenic individuals in an effort to confirm his theories of cognitive processes in
normals. Although Swinney does not specifically relate his findings to the Chapman
theory, his work offers an alternative method for testing the theory’s empirical
implications.
One of the main issues in Swinney’s research has involved the evaluation of two competing explanations as to how context influences the interpretation of words. One explanation holds that in encountering words with more than one meaning, context influences retrieval processes such that only one meaning— namely, the relevant, context-appropriate meaning—is accessed. The alternative view holds that the meaning selection process occurs after all possible word meanings are accessed. That is, contextual factors influence word comprehension independently of lexical access processes. Swinney (1979) presented evidence that supports the latter explanation, indicating that contextual and lexical processes are independent. 51
Swinney’s (1979) evidence for a separation of access and contextual processes is based on work using a cross modal lexical priming (CMLP) technique. The technique involves presenting subjects with two simultaneous tasks, namely, a sentence comprehension task and a lexical decision task. For the sentence comprehension task, subjects are presented with spoken sentences that contain ambiguous (multiple-meaning) words. They are instructed to listen to the sentences, and informed that their comprehension of them will be tested at a later point. The lexical decision task involves the visual presentation of letter strings. Subjects are asked to indicate, as quickly as possible after the presentation of each string, whether the string in question is a word or nonword. One letter string is presented for each sentence. The timing of the presentation of the string and the ambiguous word in the spoken sentence is such that the string appears either immediately after or three syllables after the ambiguous word. On some trials, the letter strings are words that are related to one of the meanings of the sentence homophones. The use of this type of string is based on work showing that lexical decisions are facilitated—that is, completed more quickly—when a target word is presented after a word that is semantically related to the target. In the CMLP task, then, the sentence homophone serves as a prime for a target item in the lexical decision task. Swinney (1984) gives the following example of a spoken sentence in the CMLP task. Asterisks are used to indicate the immediate and delayed occurrences of the word string for the lexical decision task, and the homophone is indicated by underlining.
RUMOR HAD IT THAT, FOR YEARS, THE GOVERNMENT BUILDING HAD BEEN PLAGUED WITH PROBLEMS. THE MAN WAS NOT SURPRISED WHEN HE FOUND SEVERAL SPIDERS, ROACHES, AND OTHER BUGS * IN THE COR*NER OF HIS ROOM.
In this example, the items in the lexical decision task involve either a word that is related to the intended meaning of the homophone (e.g. ANT), a word that is related to 52
the alternate meaning of the homophone (e.g. SPY), or a word that is unrelated to
either of the meanings (e.g. SEW). In the administration of the CMLP technique, the dependent variable of interest
is the time taken to respond to the target words in the lexical decision task. Swinney
(1979) reported that for trials in which the target words were presented immediately
after the ambiguous sentence words, normal individuals showed facilitated reaction
time performance in responding to targets. This result was observed regardless of the
intended meaning of the ambiguous words in the context of the sentences. These
observations suggest that immediately after an ambiguous word is encountered, the
various meanings of the word are accessed. As a result, these meanings serve as effective primes in the lexical decision task. Performance in the delayed condition, in
which the targets were presented several syllables after the occurrence of the
ambiguous words, was somewhat different from that of the immediate-presentation
condition. In the delayed condition, a priming effect was observed only for targets that
were related to the intended meaning of the ambiguous word in the context of the
sentence. Thus, although all possible meanings of an ambiguous word are accessed
immediately after its presentation, within a short time only the relevant, context-
appropriate meaning is retained. After inappropriate meanings are lost, a semantic
priming effect is not observed for primes that are related to the inappropriate
meanings.
Swinney (1984) was interested in studying the performance of schizophrenic
individuals in a CMLP task in order to provide further support for his model of word
interpretation. He noted that many of the diagnostic symptoms of schizophrenia
involve an apparent failure to utilize context appropriately. For example, the presence
of bizarre associations in speech may be due to a failure to use context in discourse production. If, according to Swinney’s model, contextual processes operate 53 independently of lexical access and come into play after lexical access has been
completed, and if schizophrenia is characterized by specific problems involving contextual abilities along with no deficits in lexical access, then schizophrenic
individuals ought to deviate from normals in a predictable manner in their performance
on a CMLP task. As word interpretation initially involves the availability of all of the
various meanings of a word, regardless of contextual factors, schizophrenic and
normal subjects should show the same pattern of performance in the immediate- presentation condition of the CMLP task. Further, as later processes in word interpretation include the selection of the appropriate meaning based on context, the performance of schizophrenic individuals in the delayed condition ought to differ from that of normals, reflecting a relative insensitivity to context.
Swinney (1984) presented chronic schizophrenic patients with a CMLP task in which sentence comprehension items contained polarized homophones, that is, words with one very common meaning and one very rarely used meaning. The results were exactly as predicted. In the immediate-presentation condition, the schizophrenic subjects showed a facilitation in lexical decision performance for words that were related to either of the two meanings of the homographs. These results indicate that schizophrenic individuals, like normals, have access to the various meanings of polysemous words at an early point in the word interpretation process. Swinney also reported that in the delayed-presentation condition, the schizophrenic subjects showed a semantic priming effect for lexical decision targets that were related to the most commonly used meaning of the ambiguous words, regardless of the intended meaning of the ambiguous words in the context of the sentence. In contrast, normal individuals showed a facilitation in lexical decision performance for targets that were related to context-appropriate meanings of homographs. Hence, in later stages of word interpre tation, normal people retain the most appropriate sense of a homophone using context, 54 whereas schizophrenic persons do not appear to use context in the selection process.
Instead, schizophrenic individuals have access to the various meanings of homographs, but invariably retain the most common meaning, irrespective of context.
The notion that lexical access processes are intact in schizophrenia is consistent with the finding that, under certain conditions, schizophrenic patients do show an adequate knowledge for the less commonly used meanings of words (Chapman et al. 1964; Deckner et al., 1971; Miller, 1974). Further, Swinney’s (1984) finding that meaning selection in schizophrenic subjects involves the retention of the most frequently used aspects of words is consistent with Chapman’s observation that schizophrenic individuals are biased toward normatively high-frequency responses.
Finally, both Swinney and Chapman have produced evidence indicating that word interpretation in schizophrenia is more heavily influenced by normal response biases than by context Swinney’s work, then, which utilizes a methodology that is quite different from the procedures developed by Chapman, provides a rather convincing case in support of Chapman’s theory.
Another study on semantic priming in schizophrenia was completed recently by Kwapil, Hegley, Chapman, and Chapman (1990). Kwapil et al. (1990) utilized a lexical decision task similar to one that was developed by Meyer and Schvaneveldt
(1971) in the study of semantic processing in normals. The task involves presenting subjects with two consecutive letter strings in rapid succession and asking them to identify, as quickly as possible, whether the second string is a word or nonword. As discussed earlier, normal individuals show a facilitation in reaction time performance when the first word is semantically related to the second word than when the two words are unrelated. These results are generally regarded as support for the notion that the processing of a given word is associated with an automatic activation of related words (Schvaneveldt & Meyer, 1973). Maher (1983) hypothesized that associative 55
intrusions in schizophrenic speech are due to a heightened spreading of activation.
According to Maher’s formulation on schizophrenic cognition, the priming of a word
is accompanied by an abnormally high activation of related words, which, although
contextually inappropriate, subsequently intrude into speech. Based on Maher’s
hypothesis (1983), Kwapil et al. (1990) predicted that schizophrenic patients would
show a heightened facilitation effect in a semantic priming task. Kwapil et al. (1990)
adapted Schvaneveldt and Meyer’s (1973) lexical decision task for use with clinical populations,7 and presented it to schizophrenic, bipolar, and normal individuals. As predicted, they found that schizophrenic subjects showed a greater facilitation effect
than bipolar subjects and normals. Hence, access to word meanings in schizophrenia
appears to be accompanied by an overactivation of related words meanings. Although
Kwapil et al. (1990) do not specifically relate their findings to the response bias theory, the results suggest a possible mechanism that may account for an excessive yielding to normal response biases in the processing of word meanings. An over activation of semantic associates would appear to explain schizophrenic patients’ inability to utilize context effectively in the processing of words meanings. Such an overactivation would also appear to account for schizophrenic patients’ tendency to substitute normatively high-frequency responses for contextually appropriate ones.
Summary
A comprehensive review of follow-up work on Chapman’s theory suggests that cognitive deficits in schizophrenia may be characterized by an accentuation of normal response biases and an insensitivity to context. The tentativeness of this conclusion comes from the several unanswered questions that have been raised in follow-up studies. On the one hand, most of the replication studies have produced results that conform to the theory’s prediction. Additionally, several studies that employ methodologies other than those used by Chapman have also provided confirming 56
evidence. A few studies have offered some rather convincing support for the theory, by virtue of their sophisticated methodologies (e.g., Swinney’s (1984) lexical priming
study) or their design (e.g., Callahan et al.’s (1986) study of blood relatives of schizophrenic patients). However, there is some suggestion that nonschizophrenic individuals may exhibit the same deficits that Chapman has ascribed to schizophrenia.
Further, as several investigators have noted (Klinka et al., 1976; Saccuzzo et al., 1980), not all schizophrenic individuals exhibit Chapman-type deficits. This observation is consistent with reports that clinically, not all schizophrenic patients exhibit thought disorder symptoms. Although chronicity of illness and length of hospitalization may be important variables in the manifestation of Chapman-type deficits, these variables have not been studied in any comprehensive way. And although the Chapman theory was designed to explain thought disorder in schizophrenia, it is possible that an exaggerated normal response bias is associated with thought disorder per se, or with certain types of thought disorder. Yet, follow-up studies have customarily compared schizophrenic subjects with control subjects, assuming that all schizophrenic subjects are thought-disordered and all nonschizophrenic subjects are not.
The emerging picture, then, is one in which the study of Chapman’s theory has produced some promising evidence along with some bothersome unresolved issues.
Such a state of affairs is typical in many areas of schizophrenia research. Neale and
Oltmanns (1980) note that compared to other early work on schizophrenia, the
Chapman studies are distinguished in terms of the reliability of their findings.
Although experimentation dealing directly with the Chapman theory has decreased in recent years, Chapman’s attention to research methodology and design continues to serve as a model for contemporary schizophrenia research. 57
Chapman’s Theory Revisited: Further Considerations
Like many early theories of schizophrenia, the Chapman theory was formulated using concepts taken from the dominant theoretical approaches of its day.
Concepts such as response biases, meaning responses, and associative responses have clear ties to stimulus-response theory and to early versions of verbal learning theory. The field of cognitive psychology, which was still in its infancy, had little influence on the early schizophrenia theories. With the rise in cognitive psychology, however, a new theoretical perspective for the study of human behavior emerged as a dominant paradigm in experimental psychology. Although cognitive psychologists incorporated many elements of previous theoretical approaches into their work, they also introduced a new set of assumptions, methods, and concepts, and defined new subject domains. As schizophrenia researchers in the 1970’s began to examine the cognitive psychology literature for ways to further understand schizophrenic deficits, they encountered a field that had little in common with existing theories of thought disorder. This may explain why the Chapman theory, along with the other early theories of schizophrenia, have been quietly set aside, and why current research on schizophrenic thought disorder tends to be conducted wholly from a cognitive psychology perspective.
A reexamination of the Chapman theory, however, suggests that the issues it deals with have received considerable attention in the information-processing literature. Specifically, work on contextual processes in normal persons may be useful in evaluating the proposition that schizophrenia involves an impaired ability to use context in the processing of word meanings. As will be discussed below, context has an all-important effect on the manner in which normal individuals process verbal material. Contextual cues guide the way that normal persons form internal 58
representations of verbal material, and can either hinder or facilitate retrieval of these
representations.
Tulving and his associates have made a number of important contributions to
the understanding of contextual processes in normal people. In a series of memory
experiments, they have demonstrated that the context in which material is encountered
determines the way that material is encoded, and that at the time of retrieval, a
reinstatement of the context in which material has been studied produces optimal
retrieval. Conversely, retrieval is greatly hindered when the context at the time of
retrieval is different from the context at the time of study. Tulving has formulated an
encoding specificity principle to account for these results: “Specific encoding
operations performed on what is perceived determine what is stored, and what is stored
determines what retrieval cues are effective in providing access to what is stored”
(Tulving & Thomson, 1973, p. 369).
Thomson and Tulving (1970, Experiment II) presented subjects with a memory
task in which to-be-remembered “target” words were studied in one context and
memory was tested in either the same or a different context. Both the study and test context were manipulated in terms of the presence or absence of “cue” words. Subjects
studied four successive lists of target words, with one study-test trial per list. Some
subjects were presented with individual target words (e.g. BLACK) and were told that their memory would be tested for all of the words presented to them. Other subjects were presented with cue-target pairs (e.g., TRAIN-BLACK), in which the cue and target words were weakly associated. Subjects who received the word pairs were told that they would be tested for their memory of the target words, but that they should study both the targets and the related cues that accompanied them.
On the first two or three trials, the context at the time of the memory tests was the same as the context at the time of study. Subjects who had studied the single-word 59
lists were simply asked, at the time of test, to recall all the words presented to them.
Subjects who had studied word pairs received memory tests containing a list of the cues from the word pairs that they had studied, and were asked to provide the target
words that corresponded to each of the cues. For example, subjects in this condition
were given the pair TRAIN-BLACK during study and presented with TRAIN-______
at the time of test. The purpose of these trials was to provide subjects with practice in
encoding the target words in relation to the cues that accompanied them.
On the third or fourth trial, the conditions of retrieval were unexpectedly changed. For these trials, the context at the time of the memory tests was different from that at the time of study. For subjects who studied word-pair lists, the memory tests contained either no cues or new cues that were strong associates to the target words (e.g. WHITE- ). For subjects who studied the single-word lists, the memory tests contained cue words that were either weak or strong associates to the targets (e.g. TRAIN- or WHITE- ).
Thomson and Tulving (1970) predicted that, for subjects who were presented with cue-target pairs, the initial practice trials would induce subjects to encode each target in a very specific manner, namely, in relation to its cue. When the conditions of retrieval are changed, however, and subjects are given cues not present at study, recall performance should suffer, despite the fact that the new retrieval cues are strongly associated with the target words. In contrast, subjects given single-word lists would be expected to encode words in a relatively broad manner, so that the presence of strong cues at the time of retrieval would not be expected to hinder performance. The presentation of strong associates at retrieval would, in fact, be expected to enhance recall.
The results of the study were exactly as predicted. Subjects who studied words in the context of cues showed lower recall when the retrieval context was changed— 60
that is, when no cues or new cues were given at retrieval—relative to their perfor
mance when the study and retrieval contexts were the same. The presence of strong
associates at retrieval, then, did not facilitate memory for targets that had previously
been studied in the context of weak associates. Subjects who studied single words
showed no decrement in performance when given cues at the time of retrieval; the
presentation of strong associates at retrieval enhanced performance, whereas the
presentation of weak associates resulted in no change in recall.
Tulving has obtained comparable results in other studies using somewhat different methodologies (e.g. Tulving & Thomson, 1971,1973; Watkins & Tulving,
1975). In all of the studies, memory performance has been affected by the similarity
between the context at study and the context at retrieval. Watkins and Tulving (1975), for example, gave subjects successive lists of weakly associated cue-target pairs. As in the Thomson and Tulving (1970) study discussed above, the memory tests in the first two trials contained the cue words that were presented at study. On the third trial, subjects received a recognition test followed by a recall test. The recognition test contained the target words, without their cues, intermixed with distractor items. The recall test contained a listing of the cue words, and subjects were asked to name the corresponding target words. Watkins and Tulving (1975) reported that subjects some times failed to recognize words that they were subsequently able to recall. This recognition failure of recallable words may seem counterintuitive in that recognition tasks involve the identification of the to-be-remembered information in an experimenter-supplied list, whereas recall tasks provide no such prompts. Yet, subjects will recall words that they cannot recognize when the recall test reinstates the context at encoding, and the recognition test provides a change in context.
A consideration of these results in light of Chapman’s theory of schizophrenic thought disorder leads to some interesting predictions. If schizophrenic individuals 61
show an accentuation of normal response biases and are relatively insensitive to
context in processing verbal information, then they should behave differently from normals in an encoding specificity paradigm. Chapman & Chapman’s (1973a)
contention that schizophrenic persons are excessively drawn to normal response biases
suggests that they tend to encode information in an invariant fashion rather than in
relation to contextual cues, as do normals. Thus, for example, the finding (Chapman,
Chapman, & Miller, 1964) that schizophrenic individuals show an excessive tendency to interpret double-meaning words in their most commonly used sense implies that
they do not use contextual cues as well as normals to mediate encoding processes.
Similarly, the observation that the performance of schizophrenic patients in a
vocabulary task involves an exaggerated tendency to respond with semantic associates
(Rattan & Chapman, 1973) may also be explained by an inflexible encoding strategy, in which contextually appropriate responses are replaced by associatively related ones.
Tulving has shown that, in remembering verbal material, a change in context between study and retrieval results in deteriorated memory performance in normals. If schizophrenic individuals do not use context to the same extent as normals in encoding words, then they should be relatively unaffected by a change in context in a Tulving- type task. In the present investigation, schizophrenic and control subjects were asked to study three successive word lists composed of weakly associated cue-target pairs.
On the first two trials, the context at the time of retrieval was the same as that at the time of study. That is, the cues present at the time of study were also present at the time of test. On the third trial, the context at retrieval was different from that at study.
Instead of presenting subjects with list cues, subjects were given new cues that were strong associates to the target words. Finally, a fourth study-test trial was administered involving the presentation of single words for study followed by memory tests in which strong associates served as retrieval cues. 62
As was the case in the Thomson and Tulving (1970) study, the first two trials in the present investigation were designed to induce the normal subjects to encode the targets in relation to the cue words. When memory was tested in the third trial, and
normal subjects were given new cues, performance was expected to suffer relative to
that of the initial trials, as it did for Thomson and Tulving’s (1970) subjects. Normal
subjects were thus expected to exhibit better memory performance on the first two
trials than on the third trial. On the other hand, schizophrenic subjects were hypothe
sized to encode target words with less regard than normal subjects to the cue words. Instead, they were expected to encode information in a relatively broad form, relying
on the strong-meaning aspects of words rather than on meaning aspects derived from
context-biasing cues. Thus, on the first two trials, schizophrenic individuals were
expected to demonstrate poorer memory performance than normals, as the presentation
of weak cues at retrieval should not facilitate memory for items stored in a broad fashion to the same extent as it presumably does for items stored in relation to the cues. On the third trial, however, the presentation of strong-associate cues was expected to facilitate schizophrenic subjects’ retrieval of these broadly encoded items.
Hence, schizophrenic individuals were predicted to show better performance on the third trial than on the first two trials. On this reasoning, schizophrenic subjects were expected to show enhanced memory performance with a change in context, whereas normal subjects were expected to show decreased performance under the same conditions.
On the fourth trial, the target words were studied without any cues and hence, no context-defining elements were present at the time of encoding. In the absence of contextual cues, both schizophrenic and normal subjects were expected to encode the to-be-remembered words in a broad fashion. The presentation of strong cues at the time of test was expected to provide optimal conditions for the recall of information 63
that had been encoded in broad form. And because both subject groups presumably
encoded the targets in this fashion, the strong cues were expected to serve as effective
retrieval prompts for both groups. For normal subjects, performance was expected to
be better on the fourth trial than on the third trial, as the latter involved a change in
context between the time of study and test. Further, normals’ performance on the fourth trial was not expected to differ from that on the first two trials, as all these trials
involved contextual conditions that presumably optimize retrieval. For schizophrenic
subjects, performance on the fourth trial was predicted to be equivalent to that on the
third trial, which also involved the presentation of strong cues at the time of test. And
schizophrenic subjects’ performance on the fourth trial, because prompted by strong
cues, was expected to be better than their performance on the first two trials, which
involved presentation of weak cues at the time of test.
The present study aims to evaluate the proposition that an exaggeration of normal response biases is associated with thought disorder. Previous work on symp toms of thought disorder has suggested that not all schizophrenic patients exhibit such symptoms, and that some nonschizophrenic patients do (e.g., Andreasen, 1979b).
Some bipolar-disordered patients, for example, have been shown to exhibit thinking disturbances that have traditionally been associated with schizophrenia. For this reason, bipolar patients were included as a psychiatric control group in this investiga tion, and the relation between memory performance and thought disorder measures was examined in all subject groups. The use of bipolar-disordered patients is also based on the rationale that patient controls should be comparable to schizophrenic subjects in terms of severity of psychiatric illness. Bipolar disorder, like schizophrenia, is commonly associated with repeated, major life disruptions, such as job loss and marital and interpersonal difficulties. At any rate, because symptoms of thought disorder are shown by some schizophrenic and some bipolar individuals, an analysis of 64
the relationship between thought disorder and performance in each patient group
provides information as to whether an accentuation of normal response biases is an
exclusive feature of schizophrenia or is present in a variety of thought-disordered psychiatric patients.
Although Chapman originally provided support for his theory in studies of
chronic schizophrenic patients, subsequent work has suggested that acute patients may
not show an accentuation of normal response biases. Hence, the performance of the
schizophrenic patients in the present investigation was examined in terms of several
chronicity variables, namely, length of illness, cumulative hospital-stay time, and
number of hospital admissions. This approach to evaluating the effect of chronicity on
performance is somewhat different from that of early schizophrenia studies. Early
studies often dichotomized schizophrenic patients into acute and chronic categories
according to the number of years of hospitalization, and then compared the perfor
mance of acute and chronic groups. However, beginning in the 1970’s, hospitalization
practices have changed dramatically in response to the trend toward deinstitutiona
lization, resulting in the maintenance of chronic schizophrenic patients in the com
munity rather than in the hospital. Thus, a years-of-hospitalization measure does not have the same meaning as it did in the 1950’s, 1960’s, and 1970’s. The practice of dichotomizing subjects into acute and chronic categories is thus not likely to produce
groups that are comparable to those of earlier studies. Therefore, the present investi gation examined the correlation between scores on the above-mentioned chronicity variables and memory performance.
As discussed in a previous section, follow-up work on Chapman’s theory has indicated that lengthy periods of institutionalization may be associated with an accentuation of normal response biases. Chapman-type deficits have been observed not only in schizophrenic inpatients, but also in nonpsychiatric medical inpatients who are hospitalized for extended periods of time. These findings leave open the possibility that lengthy hospitalization rather than schizophrenia produces heightened normal response biases. The present investigation thus examined performance in outpatient
schizophrenic subjects in order to rule out the effects of hospitalization as a source of the findings. CHAPTER n
METHOD
Subjects
Eligibility Criteria
Twenty schizophrenic individuals, 20 bipolar individuals, and 20 normal
individuals (persons without known psychopathology) served as subjects. In order to qualify for inclusion in the study, a subject, whether in either of the patient groups or
in the normal control group, had to be a native English speaker between 18 and 55
years of age with a high school or equivalent level of education. Excluded from the
study were individuals with subnormal intelligence, a history of drug or alcohol
addiction, a history of drug or alcohol abuse within two years of testing, recent major
physical illness, or a history consistent with organic brain impairment (e.g., seizures,
serious head trauma, or long periods of unconsciousness). The Vocabulary test of the
Wechsler Adult Intelligence Scale-Revised (WAIS-R; Wechsler, 1981) was used to
identify individuals with subnormal intelligence. Data from individuals who failed to
achieve a scaled score of at least 7 on the test were not included in the analyses, and
were replaced by data from new subjects.
Criteria for assignment to the patient groups included a diagnosis of either schizophrenia or bipolar depression with mania, as defined by the Research Diagnostic
Criteria (RDC; Spitzer, Endicott, & Robins, 1984), and no history of electroconvulsive therapy within two years of the time of testing. In order to qualify for inclusion in the normal control group, an individual had to have no history suggestive of serious
66 67 psychopathology. The subjects in the bipolar and normal control groups were selected so that the groups were similar to the schizophrenic group in terms of a variety of
sociodemographic and intellectual factors, including age, educational achievement, socioeconomic level, estimated intelligence, and gender and racial composition.
Identification and Recruitment of Subjects
Schizophrenic and Bipolar Subjects
The investigator used information from inpatient and outpatient medical
records to establish the patient’s RDC diagnosis. In an effort to provide some degree of
reliability for the diagnostic assessments, a patient was considered for the study only if
he or she had received a diagnosis of schizophrenia or bipolar disorder from current
treatment providers. The treatment providers’ diagnoses were formulated according to
the diagnostic criteria of the DSM-HI-R, the Diagnostic and Statistical Manual of
Mental Disorders. Third Edition, Revised (American Psychiatric Association, 1987).
Although the RDC and DSM-HI-R criteria for schizophrenia and bipolar disorder are
in many ways similar, the RDC specifies a more narrow set of criteria for the two
diagnoses, excluding borderline cases and cases with a mixture of symptoms from both
diagnoses. In reviewing patient records, the investigator thus excluded from the study patients who had been diagnosed by treatment providers as having schizophrenia or
bipolar disorder, but who did not meet the research diagnostic criteria for these disorders. As the investigator was not allowed to keep permanent records on the review process, no precise figures are available on the proportion of patients who were excluded due to diagnostic considerations. In addition, no information is available concerning the reliability of the investigator’s RDC diagnoses, as the investigator was the only researcher in this project who was permitted to examine the patient records.
Treatment staff might have been used to provide independent diagnostic assessments for the evaluation of reliability, but practical considerations precluded this alternative. 68 Staff resources were generally too limited to allow personnel to undergo training in the
use of the RDC and to perform the rather time-consuming diagnostic assessments, which required a thorough examination of virtually every record in the patient charts.
The schizophrenic and bipolar subjects were solicited from a variety of sources
in central Ohio. Slightly less than half of the patients in each group were recruited
from a large urban public mental health center. The remaining patients were recruited
in approximately equal numbers from the following facilities and programs: several
small suburban public mental health clinics, a Veterans Administration mental health
clinic, an outpatient clinic at a private psychiatric hospital, a university psychiatric
clinic, a university-sponsored schizophrenia research program, and several lay self-
help groups. Appendix A contains a listing of the specific agencies and programs from which the patient subjects were drawn.
The procedures used to identify and solicit subjects for the two patient groups varied according to the research policies of the agencies and programs that referred patients to the study. In the most typical arrangement, the investigator initially reviewed the medical records of patients who had been diagnosed by treatment providers as having either schizophrenia or bipolar disorder. The review of records provided an opportunity not only to evaluate the patient’s diagnosis, but to consider whether the patient met other eligibility criteria, such as requirements pertaining to age, education, and drug and alcohol history. At some agencies, treatment staff were available to provide additional information regarding patients whose records were sparse.
A patient who, after his or her chart review, appeared to meet the requirements for participation was contacted about the study by staff, and invited to talk with the investigator. The investigator subsequently talked with those patients who expressed an interest in learning about the study. The initial contact interview, which lasted from 15 to 30 minutes, was generally conducted by telephone. In a few instances, treatment
staff chose to personally introduce the patient to the investigator when the investigator
happened to be nearby. In these cases, the initial interview was completed in person.
During the initial interview, the investigator explained that the patient was likely to be
eligible to take part in a study on thinking and remembering. The patient was told that
if, after receiving information about the study, he or she was interested in serving as a
study participant, the investigator would need to ask the patient some questions to
determine whether he or she met all of the requirements for participation. The
investigator explained that the purpose of the study was to examine the relationship
between thinking and remembering in persons with a history of specific types of
psychiatric symptoms. The patient was told that study participants would be asked to
complete a vocabulary test, a speech task, a memory test involving numbers, and a
series of memory tests involving words. The nature of each of these tasks was
described briefly. The investigator then reviewed the rights of research participants.
The patient was told that study participants would be given $25 for their efforts. After receiving information about the study, all of the patients who were contacted expressed
an interest in participating. The investigator then conducted a semi-structured
screening interview (see Appendix B), which had several objectives. One purpose of the screening interview was to verify the accuracy of the demographic information obtained from the patient records. Another purpose was to screen for dmg and alcohol problems and for certain medical conditions, such as head injuries, neurological impairment, and serious medical illness. Patients whose records did not provide sufficient information for a definitive diagnosis were queried regarding their history of psychiatric symptoms. At the conclusion of the initial interview, patients who met the requirements for participation were invited to take part in the study. For patients who did not meet the 70 criteria for participation, the investigator discontinued the screening procedure as soon as the ineligibility became apparent. The investigator informed the patient that he or she did not meet the requirements for participation, and provided an explanation, tailored to the patient’s psychological state and level of intellectual functioning, as to why he or she was ineligible. The investigator then answered any questions that the patient had and thanked him or her for expressing an interest in the study. The investi gator occasionally encountered instances in which although a patient appeared almost certain to meet the diagnostic criteria for participation, rapport developed during the initial interview did not seem sufficient to allow asking the patient to discuss his or her psychiatric symptoms. In such instances, the investigator scheduled the patient for testing, and withheld her questions regarding symptoms until the testing session.
Normal Controls
Normal control subjects were solicited through newspaper advertisements and an employment agency. Individuals who answered the newspaper and employment- agency announcements were contacted by telephone for a 15- to 20-minute interview.
The prospective subject was told that persons were needed to serve as control subjects in a study investigating memory and thinking in psychiatric patients. The investigator explained that subjects had to meet certain minimal requirements for participation, and that subjects selected for the normal control group needed to be similar to those of the patient groups in terms of characteristics such as age and education. The investigator explained that if the respondent was interested in serving as a subject, the investigator would need to conduct a brief interview to determine whether the individual fulfilled the basic requirements for participation. The prospective subject was told that fol lowing the interviews with all of the individuals who answered the study announce ments, the investigator would schedule for testing those individuals who met the basic requirements and who were most similar to the patients in terms of age, education, and 71
the like. The investigator then provided information on the nature of the testing, the
compensation for participation, and the rights of research participants. After being
presented with this information, all of the respondents indicated an interest in serving
as research participants. The investigator then asked the prospective subjects to
respond to a series of screening questions (see Appendix B), which were the same as
those presented to the patients except that questions were added to identify and rule
out individuals with a history of psychiatric disturbance. Individuals were excluded
from the study if they had been hospitalized for psychiatric difficulties, if they had
ever been diagnosed with a mental disorder, if they had ever taken psychiatric
medications, or if they reported a history of symptoms consistent with a major
psychiatric illness. As was the case for the psychiatric patients, the investigator
terminated the interview early for individuals who did not meet the basic eligibility
requirements. Individuals who were qualified to serve in the normal control group and
who were demographically similar to subjects in the patient groups were contacted at a
later date and scheduled for testing.
Subject Characteristics
The characteristics of the subject groups are summarized in Table 1. One-way
ANOVAs indicated that the three groups did not significantly differ in terms of age,
F(2,57) = .05, g > .96, education, F(2, 57) = .62, g > .54, or WAIS-R Vocabulary
score, F(2, 57) = .45, g > .64. A chi-square test indicated that gender composition did not differ in the three groups, JC2(2, N = 60) = .53, g > .77. A Fisher’s Exact test revealed no significant group differences in terms of socioeconomic level, g = .63.
Table 1 also presents descriptive statistics for the two patient groups on several measures of hospitalization and chronicity of illness. A t-test indicated that the schizophrenic and bipolar groups did not significantly differ in terms of number of years since initial hospitalization, t(38) = .55, g > .46. Kolmogorov-Smimov tests 72 Table 1 Characteristics of the Subject Groups
Schizophrenic Bipolar Normal Measure Group Group Group
Age Mean 36.9 36.7 37.4 SD 7.1 7.4 8.1 Sex (% female) 45 55 55
Race (% black) 10 10 10
Education (years) Mean 13.7 13.7 13.2 SD 1.6 1.8 1.6
Socioeconomic levela Mode 5 5 5 Percent at mode 70 50 60 Range 3 -5 3 -5 3 -5
WAIS-R Vocabulary*1 Mean 11.3 11.0 10.7 SD 1.9 2.2 2.0
Years since initial hospitalization Mean 14.2 12.5 s d 7.6 7.3
Age at initial hospitalization Mean 22.7 24.2 SD 4.4 7.8
No. of admissions Mean 5.8 5.4 SD 4.4 3.7
Total hospital time (weeks) Median 34.5 26.0 Range 5-697 1-110
Note, n = 20 for each group. aSocial Class score based on Hollingshead’s (1957) Two-Factor Index of Social Position. Possible scores range from 1 (highest class) to 5. bScaled score on the Vocabulary test of the Wechsler Adult Intelligence Test-Revised (Wechsler, 1981). The mean of the standardization group is 10 with a standard deviation of 3. 73 revealed no significant group differences in terms of age at initial admission, = 20,
20) = .25, p > .56, number of admissions, D(n = 20,20) = .15, p > .98, or cumulative
hospital-stay time, D(n = 20, 20) = .20, p > .82.
All of the schizophrenic subjects were receiving antipsychotic medications as part of their treatment regimens. The median daily dosage, expressed in terms of chlorpromazine (CPZ) equivalence units, was 445.6 mg (range: 44.4 to 2321.4) using conversion tables by Davis (1976,1985) and Hollister and Csemansky (1990, p. 108).
Twelve of the schizophrenic subjects were also taking anticholinergic (anti-Ach) antiparkinsonian agents.^ Six schizophrenic subjects were receiving antidepressants, and three subjects were taking antianxiety medications. Of the bipolar group, 13 subjects were taking lithium; their average daily dosage was 1200 mg (SD = 273.9).
Nine of the subjects were receiving antipsychotics, with a median daily CPZ dosage of
177.8 mg (range: 25.8 to 1250.0). Three of the bipolar subjects were receiving anti-
Ach antiparkinsonian agents. Two subjects in the bipolar group were taking anti depressants, one subject was taking an antianxiety agent, three subjects were taking carbamazepine, and one subject was taking valproate. Two of the bipolar subjects were not taking any medication. Both of the medication-free bipolar subjects were under the regular care of a psychiatrist One of these patients had discontinued his medications on his own accord; the other one had discontinued her medication, under her psychia trist’s orders, after experiencing severe side-effects approximately one month before her participation in the study. Inspection of the medication data suggested that, with the exception of the two medication-free subjects, the prescriptions of patients in the present study were fairly representative of drug therapy protocols used to treat schizophrenia and bipolar disorder. 74 Thought Disorder Ratings
The Scale for the Assessment of Thought, Language, and Communication (TLC; Andreasen, 1979a, 1986) was used to evaluate thought disorder. The TLC assessment system specifies and defines 18 types of thought disorder and provides
guidelines for rating the severity of each. Many of the definitions of the various types
of thought disorder in the TLC assessment manual are identical with those that appear
in the DSM-III-R (American Psychiatric Association, 1987), which currently serves as
the chief diagnostic guide in the United States. The thought disorder ratings were based on two speech samples. A free-speech
sample was obtained by asking subjects to talk about themselves for 5 minutes; the
instructions for the free-speech sample imposed relatively minimal constraints in terms
of structuring subjects’ speech. A directed-speech sample was also obtained and
consisted of subjects’ verbatim responses to items from the WAIS-R Vocabulary test.
The free-speech and directed-speech samples were used to rate all but 2 of the 18 types
of thought disorder. Ratings of tangentiality and self-reference were based only on the directed-speech sample, as the former involves evaluating goal-directed speech, and the latter involves evaluating discourse that focuses on topics other than the speaker.
In order to evaluate the reliability of the thought disorder ratings, the speech samples were transcribed and then rated by the investigator and a judge who was blind to diagnosis. The blind rater was an undergraduate psychology major who had had some limited contact with psychiatric patients. In learning to use the assessment system, the raters reviewed the TLC assessment manual as well as information on suggested improvements to the rating system (Oltmanns, Murphy, Berenbaum, &
Dunlop, 1985). Practice assessments were then completed using the speech samples of
10 psychiatric inpatients who were not part of this study. The raters completed the first few practice assessments jointly. The remaining samples were evaluated independently with periodic conferrals to compare ratings and arrive at a consistent approach to the 75
assessment procedure. The speech samples of the subjects in this study were then
independently evaluated by the two raters. Upon completion of all of the ratings, the
raters compared their scores and, in instances of disagreement, arrived at a consensus
rating through discussion.
Although the patients, when asked to produce their free-speech samples, were
cautioned to avoid discussing matters related to their illness, they sometimes provided
significant clues that gave away their diagnostic status. Hence, the “blind” rater was
sometimes unavoidably aware that a particular speech sample had been given by a
psychiatric patient.
Memory Tests
Design The subjects were given four trials of a memory task in which sets of to-be-
remembered target words were studied in one context and memory for these words
was tested in either the same or a different context. On each trial, subjects studied a different set of 20 target words and then received a recall test followed by a recognition test. The trials varied in terms of whether or not the target words were
studied in the context of “cue” words, the strength of association between the cue and target words, and whether or not the contextual cues that were present at the time of
study were the same as those present at the time of test. Subjects were presented with the four target lists in one of four counterbalanced orders over the four study-test trials.
On any given trial, subjects assigned to different list orders studied different sets of target words, but received the same treatment with regard to the manipulation of the contextual cues. The design of the study and test conditions was patterned after the work of Tulving and associates (especially Thomson & Tulving, 1970; Tulving &
Thomson, 1973). An outline of the design, along with examples of study and test items, appears in Table 2. 76
Table 2 Format of the Memory Tasks With Examples
Test Item Format and Examples Study Item Format Trial and Examples® Recall Recognition
1 &2 Weak Cue-Target Weak Cue- Weak Cue-Target or Weak Cue-Nontarget
file-ORDER Limp- Bear-Rug away-CLOSE File- Limp-Stick limp-HAND Bear- Away-Trip bear-RUG Awav- File-Order
3 Weak Cue-Target Strong Cue- Strong Cue-Target or Strong Cue-Nontarget
moon-LOVE Count- Night-Day ruler-NUMBERS Like- Socks-Shoes lettuce-BUG Socks- Insect-Bug tie-SHOES Insect- Anger-Mad Like-Love Ill-Sick Chimney-Smoke Count-Numbers
4 Target Strong Cue- Strong Cue-Target or Strong Cue-Nontarget
THREAD Author- Town-City CITY Low- Needle-Thread BOOK Town- Long-Short HIGH Needle- Low-High Infant-Baby Friend-Foe Author-Book Gown-Dress aThe to-be-remembered (target) words appear in upper-case letters. 77
Studv-Test Trials and Conditions Trials 1 and 2 (Condition W-W). On these trials, each of the target words was presented for study accompanied by a cue word that was weakly associated with the target word, and memory was subsequently tested in the presence of the weak cues.
Trial 3 (Condition W-Sl. On this trial, weak cue-target pairs were presented for study as in the first two trials, and memory was tested in the presence of new cue words that were strong associates to the targets.
Trial 4 (Condition 0-S\ On this trial, the targets were presented for study without cues, and memory was tested in the presence of strongly associated cues.
Development of Study Lists and Memory Tests
Preparation of the study and test materials began with the construction of four lists of word triplets selected from word association norms (Bilodeau & Howell, 1965;
Cramer, 1970; Ervin-Tripp, 1970; Jenkins, 1970; Keppel & Strand, 1970; Marshall &
Cofer, 1970; Palermo & Jenkins, 1964; Postman, 1970). Each list contained 20 triplets, each of which consisted of a weak cue, a strong cue, and a target. The triplets were selected from the norms so that the characteristics of the cue and target words as well as those of the triplet lists would be as similar as possible to the characteristics of the materials used by Tulving (Thomson & Tulving, 1970; Tulving & Thomson, 1973;
Watkins & Tulving, 1975). To this end, all of the norms were consulted in the development of a set of possible candidates for the triplet lists. Those triplets that produced optimal similarity with the Tulving materials were chosen for use in this study. In both the present and Tulving studies (Thomson & Tulving, 1970; Tulving &
Thomson, 1973; Watkins & Tulving, 1975), triplets were selected so that each target word appeared in the norms as an infrequent associative response to its corresponding weak cue and as a high frequency response to its strong cue. For each of the triplet lists 78 that was used in the present study, the weak cues elicited their respective targets at an average rate of 1 % according to the word association norms; the corresponding value for the strong cues was 48%. These values were similar to those of Tulving’s triplet lists (Thomson & Tulving, 1970; Tulving & Thomson, 1973; Watkins & Tulving,
1975), which had an average association value of 1% for weak cues and 42 to 52% for strong cues. Also, as was the case for Tulving, the weak and strong cues in each of the present triplets were not associatively related to one another. The triplet words were also similar to those used by Tulving in terms of their frequency of occurrence in the
English language. The proportion of words having a Thomdike-Lorge (1944) G-Count of AA or A was approximately .9 for the target words and .7 for each type of cue word in both the present and Tulving’s studies (Tulving & Thomson, 1973; Watkins &
Tulving, 1975). The triplets comprising each of the lists appear in Appendix C.
Slides were prepared for the presentation of the study lists. As the lists were presented in a counterbalanced fashion, with one fourth of the subjects viewing each list on any given trial, two sets of slides were prepared for each list. One set contained weak cue-target pairs and was used for presenting the study items on the first three trials, and the other set contained the target words only, for presentation on Trial 4.
Each slide showed a single target or a single cue-target pair printed in black on a white background. For cue-target pairs the cue words were printed in lower-case letters above the target words, which were printed in upper-case letters. For target words appearing alone, the targets were printed in upper-case letters. The slides in each list were arranged in a random order.
Each of the memory tests was presented on a single page which listed the test items in a different random order from that used for the presentation of the study items. For each list, weak-cue and strong-cue recall and recognition tests were developed. The weak-cue versions were used on Trials 1 and 2, and the strong-cue 79
versions were used on Trials 3 and 4. The weak-cue recall tests each consisted of a list
of the 20 weak cues from a given list with a blank next to each cue. The strong-cue
recall tests contained the 2 0 corresponding strong cues listed in the same format.
The weak-cue recognition tests each consisted of a list of 20 word pairs in
which the left-hand members were the weak cues from a given list. For 10 of the pairs,
the right-hand words were the targets that corresponded to the cues; for the other 1 0
items, the right-hand members were nontarget (i.e. new) words that, according to the
word association norms, were infrequent (1 % on average) associative responses to the
cues. The strong-cue recognition tests each contained 40 pairs of words printed in two
columns. Twenty of the list items were strong cue-target pairs, and the other 20 were
strongly associated (48% on average) word pairs, or strong cue-nontarget pairs, that
did not involve any words from the triplet lists. One set of these cue-nontarget pairs, or
distractor items, was used for the Trial 3 recognition tests and another set was used for
the Trial 4 tests.
Finally, study and test materials were constructed for three practice trials, which involved presenting subjects with a total of 1 2 weak cue-target pairs and testing memory in the presence of the weak cues. The study and test items for the practice trials were prepared in a manner similar to that for the experiment proper.
Appendix D contains the study items and memory tests presented on each trial.
Procedure
The subjects were seen individually. As the experimenter had identified and recruited subjects for the study, she was not blind to subjects’ diagnoses. The duration of the experimental session ranged from 1 to 3 hours, with most subjects completing the session in 1.5 to 2 hours. The variability in the duration of the testing resulted not only from individual differences in speed of performance, but also from differences among subjects with respect to factors such as talkativeness, time required to establish 80 and maintain rapport, need for extra breaks, and inclination to dwell over individual test items.
The experimenter’s efforts to build rapport with the subject began on the initial contact interview and were continued and expanded in the testing session in order to elicit the subject’s best performance. The sequence of activities in the experimental session was designed so that tasks of relatively low cognitive demand were completed first, thus providing subjects with an opportunity to become familiar with the experimenter and the testing situation. Throughout the testing session, the experimenter maintained a supportive, nonpressuring manner in the administration of the tests, and conducted the session with a certain degree of flexibility—providing extra reassurance or allowing extra breaks—for the purpose of maximizing the subject’s motivation to put forth his or her best efforts.
Upon the subject’s arrival for the testing session, the experimenter engaged in informal conversation with the subject for several minutes, after which informed consent was obtained. For the .schizophrenic and bipolar subjects, the procedure for obtaining consent varied according to the referring facility. Typically, the consent procedure for the patient subjects began with the experimenter reading aloud a prepared statement couched in easy-to-understand language. The statement reviewed the purpose of the study, the nature of the testing, and the rights of research participants. The experimenter then asked the subject to read and sign a formal consent agreement (see Appendix E). As the normal subjects had previously received the informal statement in the mail, the experimenter merely asked them to read and sign the formal consent agreement. In obtaining consent, the experimenter encouraged all subjects to ask any questions pertaining either to the study or to their rights as participants. All subjects consented to participate, and none withdrew from the study after providing consent. 81
After obtaining informed consent, the experimenter conducted an informal 10-
to 15-minute interview during which information pertaining to various sociodemo
graphic variables, such as age and socioeconomic level, was reviewed. Although such information had been obtained previously from medical records or a telephone interview, a review of the information provided an easy, nonthreatening activity prior to the initiation of the testing. For the schizophrenic and bipolar subjects, the experimenter also obtained information on the date and length of previous psychiatric hospitalizations. Such information was used to supplement the hospitalization data gathered from patient medical records. Information on the patient subjects’ recent medication intake was also reviewed. In an effort to increase the certainty of the diagnosis for some patients, the experimenter queried them regarding specific aspects of their history of psychiatric symptoms. This process generally involved asking the subject individually-tailored questions, and lasted a few minutes.
As soon as the informal interview was completed, the experimenter administered the Vocabulary and Digit-Span tests of the WAIS-R as measures of intellectual and short-term memory functioning, respectively. Next, the free-speech sample was elicited by asking the subject to talk about himself or herself for five minutes. To facilitate generation of this sample, the experimenter suggested several topics that a subject might discuss (e.g., activities in an average day, hobbies and pastimes, growing up, work, and social activities). Subjects who were in the patient groups were asked to refrain from talking about matters related to their illness. If, during the course of speaking, a subject was silent for 15 to 20 seconds, the experimenter again suggested topics for discussion. The speech samples were audiotaped for later transcription. After finishing the speech task, the subject took a
1 0 - to 15-minute rest break, and then completed the memory tasks. 82
Administration of the Memory Tasks
Appendix F contains verbatim instructions for the memory tasks. The subject was told that slides showing words would be projected on a screen, and that memory
for the words would be tested after each set of slides was displayed. The experimenter explained that three brief practice trials would be given first and followed by four
longer trials. As a practice slide was displayed, the subject was told that each slide
would contain a clue word written in lower-case letters positioned above a memory
word in capital letters, and that the clue and memory words were in some way related
in meaning.9 The subject was asked to study the words and to try to remember as
many of the memory words as possible. In addition, the subject was instructed to pay
attention to the clue words and to note the relationship between the clue and memory
words, as this would prove helpful in completing the memory tests.
Practice Trials
The first set of practice slides contained two weak cue-target pairs and was
shown at a rate of 5 seconds per slide. After viewing the slides, the subject was given
the test materials for the weak-cue recall and recognition tests, and the experimenter
reviewed the format of the memory test materials. The subject was told that after each
set of words was shown, two memory tests would be administered. The experimenter
explained that the first memory test on each trial would contain a list of the clue words
from the slides, and that the subject would be asked to write down the memory words
next to each of their corresponding clues. For the second memory test, the subject was
told that the answer sheet would contain pairs of words. The left-hand member of each pair would be a clue word, whereas the right-hand member would be either the
memory word that corresponded to the clue or a new word that had not appeared on
the slides. The subject was asked to circle the right-hand words that were memory 83
words and to cross out the right-hand words that were new, that is, that had not
appeared in the slides.
For the second and third practice trials, the subject was presented with five
cue-target pairs in the same manner as for the first practice trial. As was the case for
the first practice trial, the recall and recognition tasks of the second and third practice
trials tested the subject’s memory for the target words in the presence of the weak
cues. Beginning with the second practice trial, the subject was asked to complete a “number task” on each of the remaining trials. The number task was administered after
the study items were shown but before the memory tests were given. The purpose of
the task was to displace the contents of short-term memory so that the ensuing memory
tests reflected memory for items stored in a more permanent form. The experimenter
explained the format of the number task before the second set of practice slides was
shown. The subject was given a sheet of paper containing random digits, and was told
that on each of the remaining trials, he or she would be asked to work on a number
task immediately following the presentation of the slides. The experimenter explained that the instructions for the task would differ from trial to trial, and that specific instructions would be announced at the beginning of each trial. The subject was further informed that the instructions for the number task would require circling or crossing out either the even or the odd numbers on the answer sheet. The experimenter instructed the subject to work as quickly as possible on the number tasks. After working on the number task for 30 seconds, the subject was told, the memory tests would be given. The experimenter then administered the second and third practice trials. The subject’s performance was closely monitored, so that any misunderstandings about the study and test procedures could be corrected before proceeding to the experimental trials. 84 Experimental Trials
As noted earlier, the four study lists of the experimental trials were presented in a counterbalanced order, with one fourth of the subjects viewing each list on any given
trial. Four list orders, corresponding to the order in which the lists were presented over
trials, were devised for use in the experiment proper. The list orders were determined
by a Latin Square arrangement (see Appendix G) of trial and list combinations. Each
subject was randomly assigned to one of the four list orders, with the restriction that
equal numbers of subjects from each diagnostic group be assigned to each of the
orders. The subject was told that he or she would be given four more study-test trials in
the same manner as before, except that 20 slides would be presented on each trial. The
first two trials of the memory task proceeded exactly as the subject had been led to
expect: the subject studied weak cue-target pairs, engaged in the 30-second number
task, and then received memory tests in which the weak cues were present.
For the third trial, after studying the weak cue-target pairs and then completing
the number task, the subject was informed that the memory tests for this trial would be
different from those of the previous trials. The subject then received recall and
recognition tests involving cue words that were strongly associated with the target
words. For the recall test, the subject was told that the answer sheets provided new clues that were related in meaning to the memory words, and was asked to write down the memory word that corresponded to each of the new clues. For the recognition test,
the subject was told that each word pair listed on the answer sheet contained words
that were related to one another, and that in some of the pairs, the right-hand member was a memory word. The subject was instructed to circle the right-hand members that were memory words and to cross out the right-hand members that were new words that had not appeared on the slides.
For the last trial, the subject was told that he or she would be presented with memory words without any clues, and that the subsequent memory tests would contain 85 clue words to assist memory for the targets. The subject was then presented with the single target words followed by the number task. As in the third trial, the memory tests contained cue words that were strong associates to the target words. For the recall test, the subject was asked to write down the memory words next to their corresponding clues. For the recognition test, the subject was asked to identify the pairs in which the right-hand word was a memory word, by circling the right-hand member if it was a memory word, and crossing it out if it was not a memory word. CHAPTERID
RESULTS
Thought Disorder Ratings
Reliability
Table 3 presents several measures of interrater consistency for each category of
thought disorder. The kappa values1 0 (J. Cohen, 1960,1968) were significant for
approximately a third of the categories. Both the weighted and unweighted kappa
values were comparable to those of other investigators (Berenbaum, Oltmanns, &
Gottesman, 1985; Oltmanns, Murphy, Berenbaum, & Dunlop, 1985) who have trained
undergraduate students in the use of Andreasen’s (1979a, 1986) TLC rating scale.
Note that kappa is zero for half of the categories, despite high rates of absolute
agreement between the two raters. The zero values are a result of the fact that virtually
all subjects received a rating of zero for these categories. Other investigators using the
TLC scale in the assessment of schizophrenic and manic inpatients have also reported
infrequent ratings other than zero—which produce zero or near-zero values of kappa—
for the same thought disorder symptoms (Berenbaum, Oltmanns, & Gottesman, 1985;
Harvey, Docherty, Serper, & Rasmussen, 1990; Neale, Oltmanns, & Harvey, 1985;
Oltmanns, Murphy, Berenbaum, & Dunlop, 1985).
For any given category of thought disorder, the finding of a significant
weighted kappa led to the inclusion of that category in subsequent analyses of the relationship between thought disorder, diagnosis, and performance on the memory
tasks. Symptoms that were not reliably rated were excluded from consideration. The
86 87
Table 3 Interrater Consistency in Judgements of Thought Disorder
Percent Unweighted Weighted Category Agreement Kappa Kappa
Poverty of speech 93 '69** .83** Poverty of content 78 29* .41* Pressure of speech 87 .67*** 89*** Distractible speech 98 0 0 Tangentiality 80 . 1 2 .38 Derailment 63 .35** 4 9 ** Incoherence 95 0 0 Illogicality 98 0 0 Clanging 1 0 0 0 0 Neologisms 1 0 0 0 0 Word approximations 97 0 0 Circumstantiality 62 .30** .44* Loss of goal 90 .46* .73* Perseveration 93 0 0 Echolalia 1 0 0 0 0 Blocking 98 0 0 Stilted speech 93 - . 0 1 -.03 Self-reference 83 .28 .34 Subjective global rating 42 .06 .37*
Note. All significance tests were one-tailed. *2<.05. **£<.01. *** 2 <.001. 88 subsequent analyses thus included the judges’ ratings for the following thought disorder categories: poverty of speech, poverty of content, pressure of speech, derailment, circumstantiality, and loss of goal.
Frequency of Thought Disorder Symptoms
Table 4 provides data on the percentage of schizophrenic, bipolar, and normal subjects who were rated as exhibiting thought disorder symptoms. The entered values are based on the consensus ratings of the two judges, and represent the percentage of subjects who obtained a rating of two or more. The possible ratings ranged from 0 to 3 for some categories and 0 to 4 for others, with a rating of 0 corresponding to the absence of the symptom, and ratings of 1 to 4 corresponding to the presence of mild, moderate, severe, and extreme symptoms, respectively. Hence for each category, the percentage values in Table 4 include subjects who were rated as having at least moderate symptoms. Note that few of the normal subjects were observed with thought disorder symptoms. Over half of the schizophrenic subjects were rated as displaying symptoms of derailment, and slightly less than half were judged as exhibiting global symptoms of thought disorder. Six other symptoms of thought disorder were each observed in a small proportion (5 to 15%) of the schizophrenic group. In contrast, for bipolar subjects, symptoms of pressure of speech, derailment, and global thought disorder were each observed in approximately a quarter of the subject group. Three other symptoms were each observed in a small proportion of the bipolar group. These results are consistent with previous reports that during periods of remission, thought disorder symptoms tend to persist more in schizophrenia than in bipolar disorder
(Andreasen & Grove, 1986). 89
Table 4 Percentage of Schizophrenic, Bipolar, and Normal Subjects Exhibiting Thought Disorder Symptoms3
Percentage of Subjects
Schizophrenic Bipolar Normal Category Group Group Group
Poverty of speech 5 0 5 Poverty of content 5 0 5 Pressure of speech 5 25 5 Distractible speech 0 5 0 Tangentiality 1 0 l l b 5 Derailment 60 30 5 Incoherence 5 0 0 Illogicality 0 0 0 Clanging 0 0 0 Neologisms 0 0 0 Word approximations 0 0 0 Circumstantiality 15 1 0 5 Loss of goal 1 0 0 0 Perseveration 0 0 0 Echolalia 0 0 0 Blocking 0 0 0 Stilted speech 0 0 5 Self-reference 0 0 b 0 Subjective global rating 45 25 5
Note, n = 20 for each group unless otherwise noted. aBased on a consensus rating of 2 or more—indicating moderate to extreme symp toms—on Andreasen’s (1986) Scale for the Assessment of Thought, Language, and Communication. bBased on 19 subjects. 90
Memory Task Performance
The recall and recognition data were evaluated separately. For each type of data, a repeated-measures Latin-square analysis of variance (ANOVA) was performed, with diagnostic group and list order serving as between group factors, and trial and list serving as repeated factors. The diagnostic group factor was comprised of three levels, corresponding to the schizophrenic, bipolar, and normal groups. The trial, list, and list order factors were each comprised of four levels. The Latin-square aspect of the analysis pertained to the treatments involving the trial, list, and list order factors. As discussed previously, the four levels of the list order factor were drawn from a counterbalanced Latin-square arrangement of trial-list combinations. The appropriate statistical model for such a design partitions the variance in a different manner from that of a standard analysis. Winer (1971, pp. 727-736) and Myers (1979, pp. 279-285) present the analytic model and computational formulas for the particular Latin-square design used here.
Recall Performance
The dependent variable for the recall tests was the number of words recalled.
Table 5 presents the means and standard deviations of the recall scores for the three subject groups across trials. The ANOVA results (see Table 6 ) revealed a significant main effect for the trial and list factors, and a significant Group X Trial interaction.
Because the significant list factor was unexpected, Tukey tests were used to compare performance in the four list conditions. The means (and standard deviations) for the four lists were 10.6 (5.5), 10.0 (5.3), 10.1 (4.4), and 12.3 (4.9). Post-hoc comparisons revealed that performance on one list (List 4, as given in Appendix C) was significantly better than on the other lists, which did not differ from one another. Note, however, that although one list was easier than the other lists, the list factor did not 91
Table 5 Mean Recall Scores and Standard Deviations for Schizophrenic, Bipolar, and Normal Subjects as a Function of Trial and Cue Condition at Study and Test
Trial and Cue Condition3
1 2 3 4 Subject Group W-W W-W W-S 0-S
Schizophrenic Group Mean 11.4 10.5 10.5 11.5 £D 5.0 4.4 4.9 4.9
Bipolar Group Mean 9.7 9.5 7.7 11.7 £D 4.8 4.3 4.9 5.2
Normal Group Mean 13.7 13.6 6.7 12.9 SD 3.7 4.3 5.3 5.6 aW-W = Weak cues present at study and test. W-S = Weak cues present at study and strong cues present at test. 0-S = No cues present at study and strong cues present at test. 92
Table 6 Analysis of Variance of Number of Words Recalled
Source______df______MS______E Between subjects
A (diagnostic group) 2 90.40 1.43 B (list order) 3 71.22 1.13 A xB 6 74.27 1.18 Error (between) 48 63.14
Within subjects
C (trial) 3 167.85 2 0 .6 6 * D (list) 3 70.89 8.73* A x C 6 57.75 7.11* A x D 6 8.78 1.08 C x D (B) 6 12.28 1.51 A x C x D (B) 1 2 5.00 0.61 Error (within) 144 8 . 1 2
*2 < -0001. 93
interact with any of the other factors. Hence, the effects of the other factors can be interpreted independently from those of the list factor.
The Diagnostic Group X Trial interaction was evaluated in light of the
differential predictions that were formulated from the response bias theory and the
encoding specificity principle. As discussed earlier, the subject groups were expected
to differ in their pattern of performance over the four study-test trials. Figure 1
provides a graphic representation of each of the three groups’ performance profiles.
For the schizophrenic and normal group, planned simple-effect comparisons
were initially carried out to determine whether each subject group exhibited the
predicted pattern of performance over trials. If the initial comparisons were consistent
with predictions, subsequent interaction comparisons (complex contrasts) were performed to ascertain whether the pattern of performance over trials for a given group
was different from that of the other subject groups.
Five comparisons were planned for the normal group. In order to control for
familywise error, the significance levels were adjusted according to the Bonferroni-
Dunn technique. The five comparisons involving the normal subjects were viewed as a
family of related contrasts. Thus, the probability level for Type I error was set at .05/5
or . 0 1 per comparison.
The first three comparisons examined whether the normal subjects in the
present investigation performed in the same manner as subjects in Tulving’s studies of
encoding specificity. Normal subjects were predicted to show facilitated memory
performance on Trials 1,2, and 4 relative to their performance on Trial 3. The
predicted performance profile over trials for the normal group is depicted in Figure 2.
Within-subject comparisons revealed that the normal groups’ recall performance conformed with predictions. The first comparison contrasted normal
subjects’ performance on Trials 1 and 2. This contrast was not significant, E(l, 16) = 94
20
18 *8 16 I 14 CO * 2 12
0
1 2 3 4 W-W W-W W-S o-s Trial and Cue Condition
Figure 1. Mean number of words recalled by schizophrenic, bipolar, and normal subjects as a function of cue condition at study and test (W-W = Weak cues present at study and test. W-S = Weak cues present at study and strong cues present at test. 0-S = No cues present at study and strong cues present at test.) cues present at study and test W-S = Weak cues present at study and strong cues present present cues and strong study at present cues = Weak W-S test and study at present cues at test. 0-S = No cues present at study and strong cues present at at test.) present cues strong and study at present cues = No 0-S test. at Figure 2. Predicted memory performance profile for the normal group. (W-W = Weak Weak = (W-W group. normal the for profile performance memory Predicted 2. Figure
Memory Performance W-W 1 Trial and Cue Condition Cue and Trial W-W 2 W-S 3 o-s 4 95 96
0.00, J2 < .95. Next, the normal subjects’ average performance on Trials 1 and 2 was contrasted with their performance on Trial 4. This comparison was also nonsignificant,
E(l, 16) = 0.77, j 2 < .40. Thirdly, normal subjects’ average performance on Trials 1,2 , and 4 was compared with their performance on Trial 3. This comparison was significant, £(1,16) = 85.31, J2 < .0001. These three findings, involving a facilitation in performance on Trials 1,2, and 4 relative to Trial 3, thus constitute a successful replication of Tulving’s work, in which similar treatments were administered.
Complex contrasts were then evaluated in order to determine whether the performance profile that was predicted for the normal group was differentially present in the normal versus the other groups. In other words, the complex contrasts tested whether the normal group differed from the other groups in terms of the difference between performance on Trial 3 and the average performance on Trials 1,2, and 4.
These contrasts revealed that the normal group differed from the schizophrenic group,
E(l, 32) = 29.57, p < .0 0 0 1 , as well as the bipolar group, E(l, 32) = 11.71, j> < 0. 0 2 , in this performance pattern. Thus, the normal subjects responded in accordance with the encoding specificity principle, and their performance was significantly different from that of the schizophrenic and bipolar subjects.
Schizophrenic subjects were predicted to show better memory performance on
Trials 3 and 4 than on Trials 1 and 2. The performance profile corresponding to these predictions is illustrated in Figure 3. As was the case for the normal group, three simple-effect comparisons, followed by two interaction contrasts, were planned to test the predictions pertaining to the schizophrenic group. Following the rationale that was used in adjusting the significance level for the normal subjects, the significance level for the 5 planned comparisons involving the schizophrenic subjects was set at .05/5 or
. 0 1 per comparison. 97
ou £ g *§
0 1
1 2 3 4 W-W W-W W-S 0-S
Trial and Cue Condition
Figure 3. Predicted memory performance profile for the schizophrenic group. (W-W = Weak cues present at study and test W-S = Weak cues present at study and strong cues present at test 0-S = No cues present at study and strong cues present at test) 98
The simple-effect comparisons indicated that the schizophrenic group did not show differential performance on Trial 1 versus Trial 2, F(l, 16) = 1.26, j> < .28, nor
on Trial 3 versus 4, E(l> 16) = 0.67, p < .43. Contrary to predictions, the schizophrenic
group did not show enhanced performance on the average of Trials 3 and 4 relative to
the average of Trials 1 and 2, F(l, 16) = 0.01, p < .94.
Consequently, post-hoc comparisons were performed in order to further
examine the schizophrenic group’s performance over the four trials. The comparisons
were conducted so that the fewest number of contrasts were performed for a
meaningful interpretation of the results. This approach allowed for more sensitive
probability levels in significance testing compared with an approach in which all possible comparisons are made. Because Trials 1 and 2 involved the same treatment
manipulations, and because schizophrenic subjects did not exhibit differential memory performance over these two trials, performance on Trials 1 and 2 was combined.
Essentially, the post-hoc comparisons contrasted performance in each treatment condition with that in every other treatment condition. Hence three comparisons were considered. One of these post-hoc comparisons—involving a contrast between Trial 3 and Trial 4—had already been completed as a planned comparison. One of the remaining post-hoc comparisons contrasted the subjects’ average performance on
Trials 1 and 2 with their performance on Trial 3. The last comparison contrasted schizophrenic subjects’ average performance on Trials 1 and 2 with their performance on Trial 4. The significance level for the post-hoc comparisons was adjusted to accommodate for the fact that only three of the five planned comparisons were completed. The collective significance level for the group of completed planned comparisons was .03. In order to maintain a significance level of .05 for the family of comparisons involving the schizophrenic subjects, the post-hoc comparisons were carried out with a collective significance level of (.05 - .03) or .02. Because two 99
comparisons were evaluated on a post-hoc basis, the significance level for each
comparison was set at .02/2 or .01. The use of the Bonferroni-Dunn technique in
setting significance levels for post-hoc tests involving repeated measures is discussed
by Myers (1979, pp. 302-303).
The results were that none of the post-hoc comparisons were significant.
Schizophrenic subjects’ average performance on Trials 1 and 2 did not differ from
their performance on Trial 3, E(l, 16) = 0.27, p < .62. Further, the group’s average performance on Trials 1 and 2 did not differ from performance on Trial 4, F(l, 16) =
0.36, p < .56. Thus, the schizophrenic subjects did not perform differentially across
trials.
No specific predictions were made for the bipolar subjects. Four simple-effect
comparisons were performed, and each comparison was evaluated at a significance
level of .05/4 or .0125. As was the case for the post-hoc comparisons involving the
schizophrenic subjects, the number of comparisons involving the bipolar subjects was kept to a minimum. The approach to evaluating the bipolar group’s performance took into account the fact that Trials 1 and 2 involved the same experimental treatments. A comparison of the recall scores of the bipolar group on Trials 1 and 2 was not significant, E(l, 16) = 0.13, p < .73. Hence, performance on these trials was combined in subsequent comparisons. The bipolar subjects’ average performance on Trials 1 and
2 did not differ from their performance on Trial 3, E(l> 16) = 2.84, p < .12. A comparison of the average performance on Trials 1 and 2 with performance on Trial 4 was barely significant, E(l> 16) = 8.23, p < 0. 1 1 1 . Finally, a comparison of performance on Trial 3 versus Trial 4 was significant, F(l, 16) = 20.73, p < .0003.
Hence, the bipolar subjects recalled more words on Trial 4 than on the other trials. 100
Recognition Performance
Subjects’ performance on the recognition tests was evaluated in terms of d'
scores. The use of d' in the assessment of recognition performance is derived from signal detection theory, which provides a conceptual framework for measuring
sensitivity and decision bias in the detection of events (Green & Swets, 1966; Macmillan & Creelman, 1991). As applied to the measurement of recognition
memory, d1 provides an index of a subject’s ability to discriminate between previously
studied, to-be-remembered (target) items and distractor (nontarget) items in a
recognition task. This application presumes that a subject responds to recognition test
items on the basis of their familiarity. Specifically, when the familiarity of a test item
exceeds a given value, the subject responds that the item is old, that is, that the item is
a to-be-remembered one. In contrast, when the familiarity of a test item falls below the
given value, the subject responds that the item is new. The familiarity distributions for
the two types of items overlap, although the mean familiarity of to-be-remembered
items is generally higher than that of distractor items. A subject’s ability to distinguish
between to-be-remembered and distractor items in a recognition task thus is a function
of the amount of overlap of the familiarity distributions. Recognition sensitivity
increases as the difference between the means of the distributions increases and the
overlap of the distributions decreases. A d' score represents the difference, expressed
in standard score units, between the mean familiarity value of to-be-remembered items
and that of distractor items. Higher values of d' correspond to greater levels of
accuracy in recognition judgments.
Table 7 presents the means and standard deviations of the d' scores for the three
subject groups across trials, and Table 8 presents the analysis of variance results. Like the analysis of the recall data, the analysis of the recognition data indicated significant effects associated with the trial factor and the Diagnosis X Trial interaction. Unlike the 101
Table 7 Mean d' Scores and Standard Deviations for Schizophrenic, Bipolar, and Normal Subjects as a Function of Trial and Cue Condition at Study and Test
Trial and Cue Condition3
1 2 3 4 Subject Group W-W W-W W-S 0-S
Schizophrenic Group Mean 2.52 2.17 1.82 2.15 sn 0.81 1.07 1 . 0 2 0.90 Bipolar Group Mean 1.67 1.73 1.23 2.13 SD 0.99 1.17 0.78 1.03
Normal Group Mean 2.64 2.65 1.77 2.65 £D 0.84 0.76 0.79 0.81 aW-W = Weak cues present at study and test. W-S = Weak cues present at study and strong cues present at test. 0-S = No cues present at study and strong cues present at test. 102
Table 8 Analysis of Variance of d' Scores
Source df MS F Between subjects
A (diagnostic group) 2 11.15 4.67* B (list order) 3 1.79 0.75 A x B 6 1.78 0.75 Error (between) 48 2.39
Within subjects
C (trial) 3 6.60 17.05** D (list) 3 0.24 0.62 A x C 6 0.85 2.19* A x D 6 0.42 1.07 C x D (B) 6 0.30 0.78 A x C x D (B) 1 2 0.25 0 . 6 6 Error (within) 144 0.39
*£<.05. **£<.0001. 103
analysis of the recall data, the recognition analysis produced a significant effect for diagnostic group. In addition, the list factor was not significant. Hence, subjects did
not show differential performance in any of the four list conditions.
The significant Diagnostic Group X Trial interaction was evaluated in the same
manner as that used in analyzing the recall results. Figure 4 provides a graphic representation of the groups’ performance profile over trials.
Simple-effect comparisons revealed that the normal subjects exhibited the same
performance pattern over trials for recognition as for recall. The normal subjects’ d1
scores on Trials 1 and 2 did not differ, E(l* 16) = 0.01, p < .93. Their average d' scores
on Trials 1 and 2 did not differ from their d' scores on Trial 4, F(l, 16) = 0.00, p < .97.
The normal subjects’ average d' performance on Trials 1,2, and 4 differed from their
d' scores on Trial 3, F(l, 16) = 38.24, p < .0001. Thus, the recognition performance
profile of the normal group was characterized by a decrease in performance on Trial 3
compared to the other trials. As was the case with normal subjects’ performance on the
recall tests, their performance on the recognition tests was consistent with predictions
derived from the encoding specificity principle.
Interaction contrasts were evaluated comparing the normal group with the other
groups in terms of the performance profile that was predicted for the normal group.
Normal subjects did not differ from either the schizophrenic group, F(l, 32) = 3.97, p < .06 (political = -01), or the bipolar group, F(l, 32) = 1.59, p < .22, in this pattern of performance. These findings are at variance with the results of the recall data analyses, which indicated that the patient groups differed from the normal controls in the extent to which they exhibited the performance pattern predicted for normals.
Although the interaction contrasts did not indicate that the performance profiles of the patient groups deviated significantly from the predicted performance profile of the normal group, the overall ANOVA contained a significant Diagnostic Group X 104
3.0 2.8 2.6 2.4 2.2 8 2.0 o 1.8 C/3 1.6 *dl 1.4 1.2 1.0 0.8 Schiz. Patients 0.6 0.4 Bipolar Patients 0.2 Normal Controls 0 i------1
1 2 3 4 W-W W-W W-S O-S
Trial and Cue Condition
Figure 4. Mean d' score of schizophrenic, bipolar, and normal subjects as a function of cue condition at study and test. (W-W = Weak cues present at study and test. W-S = Weak cues present at study and strong cues present at test. 0-S = No cues present at study and strong cues present at test.) 105
Trial interaction. Hence, the recognition performance of the schizophrenic and bipolar
groups was examined in light of the predictions for their respective groups. Originally, five comparisons were planned for the schizophrenic group, with a
Type I error rate of .05/5 or .01 per comparison. Schizophrenic subjects did not
perform differentially on Trial 1 versus Trial 2, JE(1,16) = 1.71, p.< .21. They also did
not perform differentially on Trials 3 and 4, F(l, 16) = 4.27, p < .06. Their average
performance on Trials 1 and 2 did not differ from that on Trials 3 and 4, E(l, 16) =
7.10, p < .017(^critical = -01). Although the latter comparison was close to significant,
the means were opposite in direction from the predicted pattern; the average1 dscore
on Trials 1 and 2 was 2.35 and the average d 1 score on Trials 3 and 4 was 1.99.
As the results of the simple-effect comparisons involving the schizophrenic
patients were inconsistent with predictions, the complex contrasts were not performed.
Instead, additional within-subjects comparisons were performed to evaluate the
schizophrenic group’s performance pattern over trials. These post-hoc comparisons
indicated that schizophrenic subjects’ average d' scores on Trials 1 and 2 differed from
their d' scores on Trial 3, E(l, 16) = 9.98, p < .007, but not from from their d' scores on
Trial 4, E(l, 16) = 1.70, p < .22. Thus, schizophrenic subjects exhibited a decrease in d 1
on Trial 3 relative to Trials 1 and 2, but no other comparisons were significant. Normal
subjects, in contrast, showed a decrease in performance on Trial 3 relative to all of the
other trials. Hence, schizophrenic performance was similar, but not identical, to the performance of the normal group.
The evaluation of the performance of the bipolar group revealed that this group’s d1 scores on Trial 1 did not significantly differ from those on Trial 2, E(l. 16)
= 0.13, p < .73. Scores on Trials 1 and 2 were thus combined in subsequent compari sons. The bipolar subjects’ average performance on Trials 1 and 2 was barely different from their performance on Trial 3, F(l, 16) = 7.96, p < .0123 (pcritical = .0125). The 106 group’s average d1 scores on Trials 1 and 2 did not differ from their d1 scores on Trial
4, F ( l,16) = 5.35, g < .04. Lastly, the bipolar subjects’ d' scores on Trial 3 were significantly different from those on Trial 4, E(l, 16) = 19.88, g < .0004. This pattern of performance over trials, involving a decrease in performance on Trial 3 relative to the other trials, is identical to the performance pattern of the normal subjects.
Relationship Between Thought Disorder, Diagnosis, And Memory Task Performance
Multiple regression techniques were used to evaluate the contribution of diagnosis and thought disorder to memory task performance. A series of hierarchical regression analyses were conducted to examine whether the cognitive defects identified by the Chapman theory are associated with thought disorder, schizophrenia, or schizophrenia that is accompanied by thought disorder. The analyses involved entering successive sets of variables into a multiple regression model in which memory performance was the dependent variable. The order in which the variable sets were added to the regression model was determined by their hypothesized priority. The variable sets that were entered first were comprised of factors that were potentially related to the dependent variable but were not the focus of the study. For example, age and estimated IQ were some of the variables that were introduced into the regression model first. Next, variable sets that pertained to the hypotheses under investigation were added to the regression model. Factors that were entered at these later steps included diagnostic group, rated thought disorder, and the interaction between diagnostic group and thought disorder. At each step in the analyses, the most recently entered set was examined in terms of its ability to add to the amount of explained variance in the dependent variable. The increase in variance accounted for by a last- entered set thus provides information on its relationship with the dependent variable beyond that accounted for by the previously entered variables. 107
The recall and recognition data were evaluated in separate analyses. For each type of data, three series of regression analyses were performed, with each series examining a separate dependent variable. The three dependent variables were derived by comparing the predicted trial-by-trial performance profile of the normal subjects with that of the schizophrenic subjects (see Figures 2 and 3). The first dependent variable, Difference Score 1, was defined as the subject’s memory score on Trial 3 minus the subject’s average memory score on Trials 1 and 2. As shown in Figures 2 and 3, the normal group was expected to show a decline in performance from Trials 1 and 2 to Trial 3, whereas the schizophrenic group was expected to show the reverse pattern of performance. Negative values of Difference Score 1 thus corresponded to a drop in performance on Trial 3 relative to Trials 1 and 2, and positive values corresponded to an increase.
Difference Score 2 was defined as the subject’s memory score on Trial 3 minus his or her score on Trial 4. Normal subjects were predicted to show an increase on
Trial 4 relative to Trial 3, whereas schizophrenic subjects were predicted to show no change in performance over the two trials. Thus, increasingly negative values of
Difference Score 2 corresponded to greater declines on Trial 3 relative to Trial 4.
Difference Score 3 consisted of the memory score on Trial 4 minus the average memory score on Trials 1 and 2. Normal subjects were expected to show no difference in performance over these trials, whereas schizophrenic patients were expected to show greater performance on Trial 4 relative to Trials 1 and 2. Lower values of
Difference Score 3 thus corresponded to lesser differences between Trials 1 and 2, on the one hand, and Trial 4.
In both the recall and recognition analyses, the first set of variables entered in the regression model pertained to sociodemographic and intellectual characteristics of subjects. Age, gender, education, estimated IQ, and socioeconomic level comprised the 108 first set. The gender and socioeconomic level factors were qualitative in nature and hence, dummy coding was used for their representation in the regression model. By including this set of variables in the first stage of a hierarchical analysis, subsequently entered variables could be evaluated in terms of the proportion of explained variance accounted for beyond the variance accounted for by the sociodemographic and intellectual characteristics of subjects.
In the recall analyses only, the second set of variables that was added to the regression model took into account the finding, discussed in the previous section, that one list was easier than the others. However, the inclusion of the list factor as a predictor variable in the regression analyses was logically impossible, due to the nature of the difference scores and the manner in which the list conditions were administered.
The difference scores were based on performance across trials, whereas the list factor corresponded to the particular list that a subject received on any given trial. The list order factor, which corresponded to the order in which the lists were administered over trials, was chosen to substitute for the list factor as a predictor variable. Although the list and list order factors designated different experimental manipulations, the inclusion of list order as a predictor variable in the analyses provided a means of partially representing the list factor component of the design. Because the list order factor was a nominal variable, dummy coding was used for its representation in the regression model.
Next, in both the recall and recognition analyses, the variables relating to diagnosis, rated thought disorder, and the interaction between diagnosis and thought disorder were added to the model. Because the Chapman theory does not specifically differentiate between diagnosis and thought disorder in explaining an accentuation of normal response biases, separate analyses were conducted in which the order of entry of the two variables was reversed. In one analysis (Regression 1), the diagnosis factor 109
was entered first, followed by the thought disorder factor and then the interaction
factor. In another analysis (Regression 2), the thought disorder factor was entered first,
followed by the diagnosis factor and then the interaction term.
The thought disorder variable was based on a single calculated score that has been recommended by Andreasen (1986). The derivation of the score reflects the fact
that clinicians and researchers consider some thought disorder symptoms to be more
pathological than other symptoms. Andreasen suggested that the ratings of the more
pathological symptoms be multiplied by two, and that the ratings of the less
pathological symptoms be multiplied by one. A measure of global thought disorder is
then obtained by summing the resultant scores. In the present study, the consensus
ratings of the two judges were used to calculate subjects’ global thought disorder scores. In addition, a given thought disorder symptom from the TLC scale was
included in the calculation of global thought disorder only if the symptom was rated by
the two judges with adequate reliability. Of the symptoms that were reliably rated by
the judges, those that were more pathological included poverty of speech, poverty of
content, pressure of speech, and derailment; the less pathological symptoms were
circumstantiality and loss of goal.
Recall Performance
Tables 9,10, and 11 summarize the results of the recall analyses. In each table,
the upper portion contains the results for the analyses in which the diagnosis factor
was added to the model before the thought disorder factor (Regression 1), and the
lower portion contains the results when the order of entry of these two variables was reversed (Regression 2). Note that in all of the recall analyses, the first variable set,
consisting of sociodemographic and intellectual factors, did not account for a
significant portion of variance in the dependent variables. The addition of the list order factor was associated with a significant increase in explained variance for only one of 110
Table 9 Hierarchical Multiple Regression of Recall Difference Score la
Cumulative Increase F df Variable set entered R2 inR 2
Regression 1
Sociodemographic and intellectual characteristics .06 .06 0.53 6 ,53 List order .18 . 1 2 2.41 3, 50 Diagnosis .48 .30 14.09** 2,48 Thought disorder rating .49 . 0 1 0.59 1,47 Diagnosis x thought disorder .51 . 0 2 1.14 2,45
Regression 2
Sociodemographic and intellectual characteristics .06 .06 0.53 6 , 53 List order .18 . 1 2 2.41 3, 50 Thought disorder rating .30 . 1 2 8.28* 1,49 Diagnosis .49 .19 8.81** 2,47 Thought disorder x diagnosis .51 . 0 2 1.14 2,45
Note. F and df values refer to significance testing for the increase in R2. aRecall Difference Score 1 equals the number of words recalled on Trial 3 minus the average number of words recalled on Trials 1 and 2. *ji<.01. **p< .001. Ill
Table 10 Hierarchical Multiple Regression of Recall Difference Score 2a
Cumulative Increase Variable set entered R2 inR 2 E df
Regression 1 Sociodemographic and intellectual characteristics .09 .09 0.83 6 , 53 List order .18 .09 1.92 3, 50 Diagnosis .32 .14 4.97* 2,48 Thought disorder rating .32 . 0 0 0.30 1,47 Diagnosis x thought disorder .35 . 0 2 0.72 2,45
Regression 2
Sociodemographic and intellectual characteristics .09 .09 0.83 6 , 53 List order .18 .09 1.92 3, 50 Thought disorder rating . 2 0 . 0 2 1 . 0 2 1,49 Diagnosis .32 .13 4.47* 2,47 Thought disorder x diagnosis .35 . 0 2 0.72 2,45
Note. F and df values refer to significance testing for the increase in R2. aRecall Difference Score 2 equals the number of words recalled on Trial 3 minus the number of words recalled on Trial 4. *£ < .05. 112
Table 11 Hierarchical Multiple Regression of Recall Difference Score 3a
Cumulative Increase F df Variable set entered R2 inR 2
Regression 1
Sociodemographic and intellectual characteristics .09 .09 0.84 6 , 53 List order .36 .27 6.94** 3, 50 Diagnosis .43 .08 3.32* 2,48 Thought disorder rating .46 .03 2.76 1,47 Diagnosis x thought disorder .48 . 0 1 0.64 2,45
Regression 2
Sociodemographic and intellectual characteristics .09 .09 0.84 6 , 53 List order .36 .27 6.94** 3, 50 Thought disorder rating .42 .07 5.64* 1,49 Diagnosis .46 .04 1.90 2,47 Thought disorder x diagnosis .48 . 0 1 0.64 2,45
Note. F and df values refer to significance testing for the increase in R2. aRecall Difference Score 3 equals the number of words recalled on Trial 4 minus the average number of words recalled on Trials 1 and 2. *2<.05. **£<.001. 113
the dependent variables, namely, Difference Score 3. In all of the analyses, the entry of
the Diagnosis X Thought Disorder interaction did not produce an increase in explained
variance.
The relative contribution of the diagnosis and thought disorder factors in their
association with memory performance can be examined by comparing the Regression
1 and Regression 2 portions of the tables. Consider first the recall analyses in which
the diagnosis factor was entered before the thought disorder factor (Regression 1). For
all of the dependent variables, the inclusion of the diagnosis factor resulted in a
significant increase in explained variance. The subsequent entry of the thought
disorder factor did not produce any added variance beyond that which was accounted
for by the previously entered diagnosis and other factors. Now consider the regression
analyses in which the thought disorder factor was entered before the diagnosis factor
(Regression 2). The inclusion of the thought disorder factor resulted in a significant
increase in variance for Difference Scores 1 and 3, but not for Difference Score 2. The
subsequent entry of the diagnosis factor was accompanied by a significant increase in
variance for Difference Scores 1 and 2, but not for Difference Score 3. A comparison
of the Regression 1 and Regression 2 results of the recall analyses suggests that the diagnosis factor more consistently related to the dependent variables than did the thought disorder factor.
Recognition Performance
Tables 12,13, and 14 summarize the results of the recognition analyses. As was the case for the recall analyses, the variable set comprised of the sociodemo graphic and intellectual factors did not account for significant amounts of explained variance in any of the recognition analyses. In addition, the entry of the Diagnosis X
Thought Disorder interaction did not produce any significant increase in explained variance in the recognition analyses. 114
Table 12 Hierarchical Multiple Regression of Recognition Difference Score la
Cumulative Increase Variable set entered R2 inR 2 F df
Regression 1
Sociodemographic and intellectual characteristics .15 .15 1.58 6 , 53 Diagnosis .22 .07 2.39 2,51 Thought disorder rating . 2 2 . 0 0 0 . 0 2 1,50 Diagnosis x thought disorder .25 .03 0.95 2,48
Regression 2
Sociodemographic and intellectual characteristics .15 .15 1.58 6 , 53 Thought disorder rating .16 . 0 1 0.75 1,52 Diagnosis . 2 2 .06 1.96 2, 50 Thought disorder x diagnosis .25 .03 0.95 2, 48
Note. F and df values refer to significance testing for the increase in R2. Recognition Difference Score 1 equals d 1 on Trial 3 minus the average d 1 on Trials 1 and 2 . 115
Table 13 Hierarchical Multiple Regression of Recognition Difference Score 2a
Cumulative Increase Variable set entered R2 inR 2 F df
Regression 1
Sociodemographic and intellectual characteristics .16 .16 1.69 6 ,53 Diagnosis .26 . 1 0 3.54* 2,51 Thought disorder rating .34 .08 6.08* 1,50 Diagnosis x thought disorder .38 .04 1.39 2,48
Regression 2
Sociodemographic and intellectual characteristics .16 .16 1.69 6 ,53 Thought disorder rating .18 . 0 1 0.93 1,52 Diagnosis .34 .17 6.38** 2,50 Thought disorder x diagnosis .38 .04 1.39 2,48
Note. F and df values refer to significance testing for the increase in R2. Recognition Difference Score 2 equals d' on Trial 3 minus d 1 on Trial 4. *12 < .05. **p < .01. 116
Table 14 Hierarchical Multiple Regression of Recognition Difference Score 3a
Cumulative Increase Variable set entered R2 inR 2 F df
Regression 1
Sociodemographic and intellectual characteristics .14 .14 1.47 6 , 53 Diagnosis .23 .09 3.04 2,51 Thought disorder rating .31 .08 5.74* 1,50 Diagnosis x thought disorder .32 .00 0.15 2,48
Regression 2
Sociodemographic and intellectual characteristics .14 .14 1.47 6 , 53 Thought disorder rating .20 .05 3.48 1,52 Diagnosis .31 .12 4.24* 2, 50 Thought disorder x diagnosis .32 .00 0.15 2,48
Note. F and df values refer to significance testing for the increase in R2. Recognition Difference Score 3 equals d' on Trial 4 minus the average d on Trials 1 and 2 . *12 < .05. 117
Table 12 indicates that none of the variables were significantly related to
Difference Score 1.
Table 13 indicates that the diagnosis factor accounted for a significant increase
in the explained variance of Difference Score 2, regardless of whether the factor was
entered before or after the thought disorder factor. The thought disorder factor resulted
in an increase in explained variance only when the diagnosis factor was already
present in the regression model (Regression 1). An interesting finding, which is not
revealed in the table, concerns the nature of the relationship between thought disorder
and Difference Score 2. When the regression model included both the thought disorder
and diagnosis factors, thought disorder was significantly related to Difference Score 2,
but in the opposite direction from predictions. A higher rating of thought disorder was
associated with a lower value on Difference Score 2. Lower (i.e. more negative) values
on Difference Score 2, however, corresponded to a greater decrement in performance
on Trial 3 relative to Trial 4. This pattern of performance was predicted for normal
subjects, not for thought disordered or schizophrenic subjects. Hence, an increase in
thought disorder was associated with an increase in the extent to which performance approximated that predicted for normal subjects. The fact that this relationship held only when the diagnosis factor was also present in the regression model further clouds the usefulness of the thought disorder factor.
Table 14 presents the results for Difference Score 3. When the diagnosis factor was entered into the regression model before the thought disorder factor (Regression
1), only the latter factor was significant. Conversely, when the thought disorder factor was entered before the diagnosis factor (Regression 2), only the latter factor was significant. Thus, neither of the factors contributed to explained variance unless the other factor was also included in the regression model. 118
Results of the recognition analyses, considered together, did not fare as well as those of the recall analyses. The diagnosis and thought disorder factors each contributed to the explained recognition variance in some of the analyses, but not in others. Neither of the factors were related to Difference Score 1. For Difference Score
2 , thought disorder was related to the dependent variable in the opposite direction from predictions. And for Difference Score 3, neither variable contributed to the explained variance unless the other variable was also included in the regression model.
Chronicity and Memory Performance in Schizophrenic Subjects
Correlation analyses were conducted to evaluate the relationship between chronicity and memory performance in schizophrenic subjects. Three measures of chronicity were examined: years since initial hospitalization, number of hospitaliza tions, and total number of weeks hospitalized. The memory performance measures consisted of the three difference scores and a score that was based on the sum total of a subject’s performance across all trials. For the recall data, the summed score was the total number of words recalled on the four trials. For the recognition data, the summed score was the total d score across trials. The Bonferroni-Dunn technique was used in setting critical significance levels. For each chronicity measure, four correlations were computed involving recall measures. Thus the critical significance levels were set at
.05/4 or .0125 per correlation. The significance levels for the recognition analyses were adjusted in the same fashion, resulting in a critical probability level of .0125 per correlation.
Table 15 presents the results. None of the correlations were significant at a
.0125 significance level. The adjustment in critical probability levels had no impact on the identification of significant correlations, as none of the correlations were significant at a .05 level. 119
Table 15 Correlations of Chronicity Measures with Memory Performance Scores in Schizophrenic Subjects
Yrs. Since Total Initial No. Hospital Memory Task Score Hospitalization Admissions Time (wks.)
Recall Tests
Total no. words recalled -.25 . 0 0 -.14 Difference Score 1 .04 - . 0 1 .16 Difference Score 2 . 1 0 .15 .29 Difference Score 3 -.07 -.19 -.17
Recognition Tests
Total d' score - . 2 2 . 1 1 - . 0 2 Difference Score 1 . 2 2 .32 .05 Difference Score 2 .09 .23 .14 Difference Score 3 .15 . 1 0 -.09
Note. M = 20. 120
Relationship Between Medication and Memory Performance
For both the schizophrenic and bipolar subjects, the relationship between
medication and memory performance was examined using correlation analyses. For
schizophrenic subjects, two medication measures were examined. One measure was the dosage of subjects’ neuroleptic medication, expressed in chlorpromazine
equivalence units. The other measure represented whether or not a subject was taking
an anticholinergic antiparkinsonian agent for side effects of the neuroleptic
medication. For the bipolar group, one medication measure was examined, namely,
subjects’ lithium dosage. The memory performance measures consisted of the recall
and recognition difference scores and the summed across-trials scores. As with the
chronicity analyses, the Bonferroni-Dunn technique was used to establish critical
significance levels. The probability levels were thus set at .05/4 or .0125 per correlation.
Table 16 presents the correlations. None of them were significant at a .0125 probability level. For schizophrenic subjects, the correlation between neuroleptic dosage and Recall Difference Score 3 would have been significant if a .05 level of significance had been adopted. For bipolar subjects, the correlation between lithium dosage and Recall Difference Score 3 would have been significant if the critical level had been set at .03. 121
Table 16 Correlations of Medication Measures with Memory Performance Scores in Schizophrenic and Bipolar Subjects
Bipolar Schizophrenic Group3 Groups
Neuroleptic Antiparkinsonian Lithium Memory Task Score Dosage Agent Usagec Dosage
Recall Tests
Total no. words recalled -.09 .08 -.04 Difference Score 1 .15 .07 -.36 Difference Score 2 -.25 -.31 .26 Difference Score 3 .45 .43 -.61
Recognition Tests
Total d' score -.19 -.10 . 0 1 Difference Score 1 . 1 1 - . 2 2 -.15 Difference Score 2 -.25 -.42 .33 Difference Score 3 .38 .21 -.37 aNote. N = 20. t>N = 13. The analyses included only those subjects who were taking lithium. cScored dichotomously, according to whether or not the subject was taking an antiparkinsonian agent. CHAPTER IV
DISCUSSION
Thought Disorder Ratings
The reliability values for rating thought disorder symptoms using Andreasen’s
(1979a, 1986) TLC scale were comparable to those of other studies (Berenbaum,
Oltmanns, & Gottesman, 1985; Oltmanns, Murphy, Berenbaum, & Dunlop, 1985) in which nonprofessional judges served as raters. For several thought disorder categories, the reliability values equaled or exceeded those reported by Andreasen (1979a). As in a number of other studies (Berenbaum, Oltmanns, & Gottesman, 1985; Harvey, Docherty, Serper, & Rasmussen, 1990; Neale, Oltmanns, & Harvey, 1985; Oltmanns,
Murphy, Berenbaum, & Dunlop, 1985), some thought disorder symptoms were observed infrequently in the present study, resulting in attenuated reliability values.
The inclusion of outpatient subjects, who generally show fewer thought disorder symptoms than inpatients, is also likely to have contributed to a reduction in reliability values. The instructions that were used in eliciting the free-speech samples may also have influenced the interrater reliability of thought disorder judgements. Asked to talk about themselves, subjects sometimes began discussing a particular topic related to themselves but then drifted from one self-relevant topic to another. In such instances, the raters often had difficulty distinguishing whether the subject was appropriately responding to instructions to talk about himself or herself or whether the subject exhibited symptoms of derailment or loss of goal. Use of open-ended questions
122 123 concerning specific topics, rather than broad-based instructions, might produce speech samples that are easier to rate for thought disorder and that would yield higher reliability values.
A related issue has to do with whether the speech task instructions influenced the manifestation of thought disorder symptoms. The psychiatric subjects appeared to exhibit more speech disturbance in responding to the relatively unconstrained demands of the free-speech task than in responding to the circumscribed demands of the directed-speech task. This observation is consistent with traditional clinical wisdom that schizophrenic persons exhibit more deviant behavior in situations that are relatively lacking in structure. Andreasen (1979a, 1986), who constructed the TLC scale for use in clinical as well as research settings, offers few specific guidelines for eliciting speech samples. Studies that examine thought disorder symptoms have used a variety of methods for obtaining discourse samples, including the method used here of asking subjects to talk about themselves (e.g. Andreasen & Grove, 1986). Other methods have involved the use of structured psychiatric interviews (Oltmanns,
Murphy, Berenbaum, & Dunlop, 1985); open-topic nonclinical interviews (Harvey &
Serper, 1990; Wielgus & Harvey, 1988); interviews on experiences and interests
(Andreasen, 1979a); and combinations of these assorted methods (Andreasen, 1979b).
The implicit assumption in the use of these various procedures is that the presence and amount of thought disorder in speech is unrelated to the conditions under which speech is sampled. Although such an assumption may by accurate, it remains untested.
The observation that a number of thought disorder symptoms occur infrequently casts doubt on whether study of such symptoms will substantially contribute to the understanding of schizophrenia. Symptoms such as clanging and neologisms have traditionally been cited as classic symptoms of schizophrenia, but are observed in only a small percentage of people who are diagnosed with the disorder. 124
The inclusion of rarely occurring symptoms in scales of thought disorder may have their greatest importance in situations in which a measure of global thought disorder is desired.
Memory Task Performance
In both the recall and recognition tasks, normal subjects performed in accordance with the encoding specificity principle. The normal group exhibited a decline in performance in the W-S condition relative to the W-W and 0-S conditions.
Thus, normal subjects showed poorer memory when the contextual conditions that were present at the time of retrieval did not reinstate the contextual conditions that had been present at the time of study. Hence, the experimental tasks and manipulations of the present investigation were effective in replicating the findings of Tulving and his colleagues (Thomson & Tulving, 1970; Tulving & Thomson, 1971,1973; Watkins &
Tulving, 1975). In addition, the present findings provide evidence for the encoding specificity principle in a sample of subjects who differ substantially from those of
Tulving in terms of age, socioeconomic level, intelligence, and education.
Both the schizophrenic and bipolar groups exhibited different patterns of performance on the recall and recognition tests. Moreover, the relationship between performance and thought disorder was different for the two types of memory tests. In discussing these findings, the recall and recognition tasks will first be considered separately. For each type of task, the role of diagnosis and thought disorder in explaining the results will be explored. Additionally, an attempt will be made to integrate the findings for the two types of tasks.
Recall Performance
The schizophrenic subjects’ recall performance did not vary across manipula tions in cue condition at study and test. For these subjects, the presentation of strong 125
cues at retrieval was not associated with greater levels of recall than was the presenta
tion of weak cues. On trials in which weak cues accompanied the targets at study, the
presentation of weak cues at retrieval did not produce differential recall compared to
the presentation of strong cues. And on trials in which strong cues served as retrieval
prompts, performance was unrelated to whether subjects had studied the targets in the
presence of weak cues or of no cues. The schizophrenic group’s insensitivity to cue
conditions at study and retrieval contrasted with normal subject’s context-dependent
performance. A comparison of the shapes of the performance profiles of the two
groups indicates that the schizophrenic group responded to manipulations in cue
conditions in a different fashion than did the normal subjects. That is, schizophrenic
performance was not characterized merely by quantitatively poorer levels of recall
than the normal controls. In schizophrenia research, such a finding is highly unusual.
The observation that schizophrenic patients performed as well when strong
cues were present at retrieval as when weak cues were present indicates that the two
types of cues were equally effective in aiding their recall. These findings suggest that
schizophrenic individuals encode information with respect to contextual cues and, in
addition, that they encode information in broad form, that is, as if no contextual cues
are present. Consequently, strong cues facilitate recall performance to the same extent as weak cues.
The recall performance of the schizophrenic group was not consistent with predictions. However, the finding that schizophrenic patients did not exhibit differen tial performance over trials accords with the Chapman proposition that schizophrenia is marked by an inability to use context to the same extent as normal persons. The finding that strong cues did not produce greater recall than weak cues does not accord with the proposition that schizophrenic individuals respond primarily in terms of strong-meaning responses. Instead, schizophrenic patients appear to utilize strong- and 126
weak-meaning aspects of words to the same extent. The present results, then, suggest
that schizophrenic persons rely on strong-meaning responses more than do normal
persons, but that schizophrenic persons also utilize weak-meaning responses, as do
normal persons. Chapman and Chapman might argue that this conclusion is entirely
consistent with their contention that schizophrenia is characterized by an excessive
yielding to normal responses biases. The Chapman theory does not preclude the notion
that the behavior of schizophrenic persons is sometimes appropriate and that it does not invariably involve a reliance on normal response biases. The predictions of the present study, however, were based on the hypothesis that schizophrenic behavior is
influenced more by a reliance on normal response biases than by the use of context.
This hypothesis appears to be a more strongly worded version of Chapman’s contention that schizophrenic persons rely on normal response biases to a greater extent than do normal persons.
On the recall tests, the only significant within-subject comparisons involving the bipolar group consisted of superior performance in the 0-S condition relative to the
W-W and W-S conditions. Thus, the bipolar subjects exhibited inferior performance on all trials in which the to-be-remembered information was studied in the presence of weak cues. Like the schizophrenic subjects, the bipolar subjects did not perform differentially across the first three trials. And, like the normal controls, the bipolar patients showed an increase in performance on Trial 4 relative to Trial 3. The shape of the recall performance profile of the bipolar group appears to more closely resemble that of the normal control group than that of the schizophrenic group.
In the regression analyses of trial-to-trial recall difference scores, diagnosis generally accounted for a significant amount of explained variance. In addition, for two of the three difference scores, diagnosis accounted for explained variance beyond that accounted for by thought disorder. Thought disorder was related to two of the 127 difference scores, but added little to the explained variance beyond that accounted for by diagnosis. In contrast, the interaction between diagnosis and thought disorder did not account for explained variance in any of the analyses. Hence, although Chapman formulated the response bias principle as a theory of schizophrenic thought disorder, the recall test results suggest that diagnosis rather than thought disorder may have the greater predictive power.
Recognition Performance
The analysis of variance of the recognition scores indicated that the three subject groups evidenced different performance patterns over trials. Follow-up statistical analyses revealed that the performance patterns of the groups were nearly identical. The bipolar and normal subjects exhibited highly similar performance patterns: Both groups showed a decline in the W-S condition compared to the W-W and 0-S conditions. The schizophrenic subjects did not show differential performance in the W-S and 0-S conditions, as the other groups did, but otherwise exhibited the same pattern of performance as the other groups. The shape of the performance profiles of the three groups were similar. Hierarchical regression analyses indicated that subjects’ recognition difference scores were not consistently related to diagnosis, thought disorder, or the interaction between diagnosis and thought disorder.
The recognition findings failed to conform to predictions. Schizophrenic subjects responded to manipulations in context in the same fashion as did normal controls. A consideration of the recognition results, without reference to the recall results, appears to suggest that schizophrenic individuals, like normals, encode information in relation to the contextual cues present at study. Consequently, for schizophrenic as well as normal persons, retrieval appears to be greatest in situations in which the contextual conditions at the time of test reinstate the contextual conditions at the time of study. One major problem with this conclusion, however, is that it is not 128 consistent with the recall test results. Hence, before considering the implications of the recognition results with respect to the Chapman theory, several possible explanations for the discrepancy between schizophrenic recall and recognition will be considered.
Two types of explanations will be examined; one type deals with methodological aspects of the study and the other involves a theoretical account of the inconsistency between schizophrenic recognition and recall.
Reconciling the Recall and Recognition Findings: Methodological Considerations
It is possible that the discrepancy between schizophrenic subjects’ recall and recognition performance is related to the specific types of recognition and recall tasks that were employed, or to the specific procedures that were used to administer these tasks. Consequently, one of the tasks may have provided a more valid test of the predictions than did the other. Consider first the recall tests. The only difference between the recall tests that were administered on different trials was the nature of the cues—strong or weak—that were presented to subjects. In contrast, the recognition tests varied not only in terms of the nature of the cues, but also in terms of several other factors. On the first two trials, the recognition test items consisted of weak cues paired with either their corresponding targets or nontargets. On the last two trials, the recognition task contained two types of items. One type contained a strong cue paired with a target item, whereas the other type contained strongly related word pairs, the members of which had not appeared in the study phase. Subjects also received twice as many test items on the last two trials than on the first two trials. Thus the recall tasks appear to be more comparable over trials than the recognition tasks, and consequently, may have provided a methodologically sounder test of the predictions.
Differences in the discriminating power of the recall and recognition tests are unlikely to have been a cause of the discrepant results for the two types of memory 129 tasks. As discussed previously, when populations of differing ability level are given tasks that differ in terms of discriminating power, the less able group can be expected to perform disproportionately worse on the more discriminating tasks. In the present study, normal subjects performed more poorly on Trial 3 than on the other trials; hence, the Trial 3 memory tasks were more difficult and thus more discriminating than the memory tasks of the other three trials. If the performance of the psychiatric groups had been influenced by a generalized deficit, the groups would have shown a greater decline than the normal group on Trial 3 relative to Trials 1, 2, and 4. Schizophrenic subjects clearly did not show this pattern of performance on either the recall or recognition tests. On the recall tests, schizophrenic subjects exhibited virtually invariant performance across trials. On the recognition tests, schizophrenic subjects showed a decline in performance on Trial 3 relative to Trials 1 and 2, but this drop in performance was not greater than that of the normal controls. Further, the schizo phrenic group did not show differential performance on Trial 3 versus Trial 4, as did the normal group. An examination of the recall and recognition performance of the bipolar subjects in light of the discriminating power of the tests also suggests that the pattern of these subjects’ performance was not attributable to a generalized deficit.
Reconciling the Recall and Recognition Findings: A Theoretical Explanation
As mentioned previously, the recall test results were interpreted as consistent with the idea that schizophrenic individuals encoded to-be-remembered information in two ways—in relation to contextual cues and independent of contextual cues. In con trast, on the recognition tests, the performance of schizophrenic subjects was similar to that of the normal controls. Thus, the recognition findings would seem to suggest that schizophrenic subjects, like normals, encode information with respect to contextual cues. Hence, the recognition findings, in and of themselves, appear to indicate that the 130
encoding specificity principle aptly describes memory processes in schizophrenia. The
argument to be advanced here is that the recognition test findings, like the recall test
findings, are consistent with the supposition that schizophrenic individuals encode
information in two ways. Such an interpretation can account for both the recall and
recognition results, whereas only the recognition results are consistent with an
encoding specificity explanation.
The first three trials of this experiment involved presenting subjects with to-be-
remembered words accompanied by cues. Consider the word pair COTTAGE-PEACE.
When presented with this pair for study, normal subjects presumably encoded
information about PEACE in the sense of the tranquil quietness of a summer cottage.
At the time of test, the presentation of cues that had been shown at study helped
reinstate the sense in which the to-be-remembered information had been encoded,
whereas new cues provided no such assistance. Hence, memory tests that included
COTTAGE as a retrieval cue for PEACE produced better memory performance than
those that included WAR as a cue, despite the fact that WAR is a strong associate to
PEACE. There is nothing in the encoding specificity principle that suggests that
retrieval processes in the recall and recognition tests involve different mechanisms.
For both types of tests, normals’ performance presumably is based on the extent to
which the retrieval conditions facilitated access to contextually stored information.
Schizophrenic individuals, on the other hand, are hypothesized to encode to-be-
remembered information not only in relation to contextual elements present at the time of study, but also in a form that is independent of context. When presented with
COTTAGE-PEACE, schizophrenic patients presumably encoded PEACE in the same
sense as normals did, that is, in the context of COTTAGE. In addition, schizophrenic patients also appear to have encoded PEACE in its generic sense, that is, as if it had 131
not been presented with any cues. This latter type of encoding might involve PEACE
in the sense of harmonious relations among nations.
This formulation implies that if schizophrenic subjects are given both types of
cues in a memory test, memory performance would be better than if they received only
one type of cue. An examination of the various memory tests administered over the
first three trials indicates that in some cases, the test items contained both types of
retrieval cues that presumably facilitate schizophrenic memory, whereas in other cases, only one type of cue was provided.
Consider schizophrenic subjects’ performance on the recall tests. On each of
the first three trials, memory was tested in the presence of either the cue words that had
accompanied the targets at study or new cues that were strongly associated with the
targets. Having studied COTTAGE-PEACE, subjects were given either COTTAGE or
WAR as retrieval cues at test. For schizophrenic subjects, both of these cues served as
effective retrieval prompts, as schizophrenic subjects presumably encoded the target
words in relation to the cues as well as independently of the cues. Note that according
to this formulation, the recall test cues on any given trial were used to retrieve one and
only one of the two types of encoded information—either the context-related or the
context-independent forms of the memory representations.
The recognition tests, like the recall tests, involved testing memory for targets
in the presence of either the cue words that had been present at study or new cues that
were strongly associated with the targets. However, items in the recognition tests,
unlike those in the recall tests, included the target words . 11 For schizophrenic subjects,
the target words contained in the recognition items are hypothesized to have been processed not only in relation to the cue words that accompanied them, but also independently of the cue words. After studying the pair COTTAGE-PEACE, subjects were given a recognition test containing either COTTAGE-PEACE or WAR-PEACE 132 as test items. In responding to the pair COTTAGE-PEACE, schizophrenic subjects are hypothesized to have utilized as a retrieval cue PEACE in the sense of a peaceful cottage as well as PEACE in its generic sense, that is, as harmony among nations. In responding to the pair WAR-PEACE, schizophrenic subjects are hypothesized to have utilized as a retrieval cue PEACE in its generic sense. Note that the test item WAR-
PEACE lacks any direct reference to a cottage. The two types of test items thus differed in the extent to which they provided access to the two types of encodings that schizophrenic persons presumably establish in remembering words. On trials in which the recognition tests contained the cue words present at study, test items essentially provided two types of prompts for effecting retrieval. In contrast, on trials in which the recognition tests contained new cues that were strong associates to the targets, the test items provided only one type of effective retrieval prompt. This formulation is con sistent with the observation that schizophrenic subjects’ recognition performance was higher on trials in which test items contained weak cues than on trials in which test items contained strong cues.
The post-hoc explanation offered here is consistent with Chapman’s theory of schizophrenic thought disorder. According to the present formulation, in situations in which normal individuals rely heavily on context for the representation of information in memory, schizophrenic patients utilize both context-relevant as well as context- independent information in encoding to-be-remembered material. Further, schizo phrenic individuals’ utilization of context-independent information involves the en coding of the generic, broad-based aspects of word meanings. These propositions are consistent with Chapman’s assertion that schizophrenic individuals are abnormally prone to disregard context and to respond to words in their most commonly utilized sense. 133
Relational and Individual-Item Processing
The notions presented here concerning context-dependent and context-
independent processing bear a strong resemblance to ideas presented in the general
experimental literature concerning relational and individual-item processing (Bellezza,
Cheesman, & Reddy, 1977; Einstein & Hunt, 1980; Hunt & Einstein, 1981). Relational
processing involves the encoding of information pertaining to the similarities among
to-be-remembered items, whereas individual-item processing involves the encoding of
information regarding the item-specific features of items. Evidence suggests that these
two types of processing are qualitatively different and that they result in the encoding
of different types of information in memory. Further, optimal memory performance is
observed when people process both relational and item-specific aspects of to-be-
remembered material. However, the nature of the studied material and the task
instructions given to subjects at the time of study influence the type of processing that
normal subjects utilize in representing the material in memory. For example, when
presented with a list of randomly arranged words drawn from several categories,
normal persons have been shown to rely heavily on the categorical relationships
among list items in representing the material in memory. Presumably, then, catego
rized word lists tend to evoke an encoding strategy involving the processing of the relational attributes of the material. If, at the time of study, subjects are asked to
organize the list items according to categories, their subsequent memory performance is not as high as if they are given an orienting task that involves individual-item processing, such as rating the pleasantness of each item. This observation is consistent with the notion that, in memory tasks in which the to-be-remembered material has an obvious organizational structure, instructions that involve additional processing of relational information at the time of study are not as effective as instructions that involve the processing of individual-item features. On the other hand, when subjects 134
are presented with a list of unrelated words, evidence suggests that subjects rely less
on relational processing than on the processing of item-specific features. In such a case, an orienting task at study that induces the processing of the relationships among
list items —such as asking subjects to create a story using the list words—produces higher memory performance than an orienting task that induces the processing of item-
specific features, such as having subjects rate the pleasantness of each list word. These findings have been interpreted as evidence that for to-be-remembered material that lacks a salient organizational structure, instructions that require the processing of individual-item features do not enhance memory as much as instructions that require
subjects to process the relationships among study items.
The results of the present study can readily be interpreted in terms of conceptu alizations regarding individual-item and relational processing in normal persons. In the present investigation, subjects were presented with a memory task involving the study of weakly associated word pairs. The relationship between the members of each word pair was thus not readily apparent However, the instructions for the task emphasized that the words of each pair were related in some manner in terms of meaning, and that subjects should study the words by noting the relationship between the words of each pair. Further, subjects were led to believe that the memory tests on each trial would involve the presentation of one word of each pair, and that they would be asked to demonstrate memory for the corresponding member. The memory task instructions thus effectively presented subjects with an orienting task that directed processing towards the relational aspects of to-be-remembered information.
Normal subjects in the present investigation showed evidence of having encoded information primarily in terms of the relational aspects of the word pairs: On trials in which the memory tests contained cues that had been presented at study, memory performance was better than on trials in which the tests contained new cues. 135
Presumably, in the W-W condition, the test cues provided subjects with retrieval
prompts that facilitated access to information stored in relational form. In contrast, in
the W-S condition, subjects received test cues that lacked such relational information.
And because the to-be-remembered words had been encoded in terms of their
relational aspects, the new cues did not facilitate retrieval to the same extent as the
cues that had been present at study.
Schizophrenic subjects, in contrast, appeared to process target words in a
different fashion than the normal controls. On the recall tests, schizophrenic subjects
did not exhibit differential performance across manipulations in cue condition. This
finding suggests that schizophrenic subjects processed more than one type of informa
tion in representing items in memory, and consequently, more than one type of retrieval cue was effective in producing access to stored information. The observation that strong cues facilitated retrieval as well as weak cues indicates that schizophrenic
subjects, unlike normal ones, processed item-specific and relational information to the
same extent. Further, the effectiveness of the strong cues for schizophrenic subjects suggests that such item-specific information involved a representation of the generic meanings of the to-be-remembered words. Hence, schizophrenic subjects were less able than normal subjects to modify encoding processes in response to task demands.
On the recognition tests, schizophrenic subjects exhibited differential memory performance across trials. As discussed earlier, one potential explanation of this finding holds that on some trials, the recognition test items in effect provided schizophrenic subjects (but not normal ones) with two types of cues, whereas on other trials, the recognition test items provided only one type of cue. Presumably, the recognition tests that included two types of cues provided schizophrenic subjects with both item-specific and relational retrieval prompts, whereas tests that included only one type of cue provided only an item-specific prompt. 136
Potential Mechanisms of Encoding Deviance in Schizophrenia
The data provide little evidence as to whether schizophrenic subjects’ pre sumed use of context-independent as well as context-related information was a product of a consciously controlled encoding strategy or of an automatically controlled pro cess. Semantic priming studies have produced evidence indicating that lexical access processes in schizophrenia are characterized by a heightened activation of the strong- meaning aspects of words (Swinney, 1984; Kwapil et al. 1990). Such findings suggest that schizophrenic persons may process context-independent information more easily than context-related information because lexical access processes are marked by an abnormal tendency to favor strong-meaning aspects of words. In contrast, the process ing of contextual information in many cases requires that nondominant aspects of word meanings be utilized. In such cases, lexical access processes that favor strong-meaning aspects of words do not facilitate the processing of contextual attributes. Hence, for schizophrenic individuals, the processing of contextual information may require more cognitive effort than the processing of context-independent information.
The supposition that schizophrenic encoding abnormalities are due to distur bances in automatic lexical access processes does not explain all of the available evidence. Several studies have demonstrated that engaging schizophrenic patients in orienting tasks at the time of study results in an amelioration of memory deficits.
Larsen and Fromholt (1976), for example, presented hospitalized chronic schizo phrenic patients and normal controls with a sorting task followed by a free-recall task.
Subjects were given a set of unrelated words and were asked to sort them by placing words that belonged together in the same group. Subjects were required to repeat the sorting task until they sorted the words in an identical fashion on two consecutive trials. After reaching this criterion, an unexpected free-recall task was administered.
Larsen and Fromholt (1976) found that the schizophrenic and normal groups did not 137
differ in recall performance. This finding contrasts with the well-documented
observation (Koh, 1978) that hospitalized chronic schizophrenic patients exhibit
deficient performance on recall tasks in which subjects are merely asked to study and
then recall a list of items. Larsen and Fromholt’s (1976) findings thus suggest that schizophrenic individuals’ memory difficulties can be ameliorated with instructions
that induce the processing of relational attributes of material.
Koh, Kayton, and Peterson (1976) presented hospitalized schizophrenic
patients, hospitalized nonschizophrenic psychiatric patients, and normal controls with
an orienting task in which subjects rated the pleasantness of each of a set of unrelated
words. Following the pleasantness ratings, subjects received an unexpected free-recall
task. Koh et al. (1976) found that schizophrenic patients did not differ from either
psychiatric controls or normal controls in memory performance. Hence, the comple
tion of pleasantness ratings, which presumably direct attention to individual-item
aspects of to-be-remembered material, was associated with an absence of a recall
deficit in schizophrenic patients.
The Larsen and Fromholt (1976) and Koh et al. (1976) findings suggest that
orienting tasks that promote the processing of either relational or item-specific aspects
of to-be-remembered material appear to improve schizophrenic subjects’ ability to encode and retrieve information. Further, schizophrenic individuals’ difficulties in encoding information may be due not only to abnormalities in automatic processes,
such as problems with lexical access, but also to problems in selecting efficient encoding strategies.
Recently, J. D. Cohen and Servan-Schreiber (1992) presented an ambitious formulation that holds that schizophrenic performance deficits on disparate cognitive tasks are attributable to a disturbance in the use of context. Their formulation is based on computer simulation models that incorporate both information-processing and 138 biochemical parameters of cognitive processes. The investigators built computer simulation models constructed to reflect normal individuals’ performance on two attentional tasks and a sentence interpretation task similar to that of Chapman,
Chapman, and Miller (1964). Each of the computer models contained features designed to simulate normals’ use of context in the execution of the tasks. In addition, the models included features designed to simulate the neuroanatomical and neuro- chemical mechanisms that normally are involved in the use of context. J. D. Cohen and
Servan-Schreiber (1992) demonstrated that the models aptly mirrored the performance of normal individuals. Further, when the contextual parameters of the models were altered, the performance of the models was similar to that of schizophrenic subjects.
This work thus provides an intriguing account of the manner in which disturbances in the use of context produce performance deficits on a variety of cognitive tasks.
J. D. Cohen and Servan-Schreiber’s (1992) theory not only is consistent with Chapman and Chapman’s (1973a) assertion that cognitive dysfunction in schizophrenia is a result of problems in the use of context, but also delineates specific mechanisms, described in information-processing and biochemical terms, that explain such dysfunction. Further, the view that problems in the use of context disrupts a number of cognitive processes suggests that encoding deviance may be explainable in terms of a disruption of multiple mechanisms. This view is consistent with evidence, discussed above, that encoding deviance may result from abnormalities in both automatic pro cesses as well as consciously controlled ones.
The Role of Thought Disorder
Although the Chapman theory was intended to explain schizophrenic thought disorder, the results suggested that diagnosis may have more predictive power than thought disorder in accounting for Chapman-type deficits. The finding that thought 139 disorder was associated with recall performance, but added little to the explained variance beyond that accounted for by diagnosis, indicates that thought disorder and diagnosis shared common variance in their association with performance. This finding
suggests that thought disorder may be one aspect of an aberrant cognitive process that is fundamental to schizophrenia and that also produces anomalies in encoding. That is, both thought disorder and encoding deviance may be the end result of some higher order cognitive dysfunction that underlies schizophrenia. Note that this suggestion is consistent with J. D. Cohen and Servan-Schreiber’s (1992) contention that a given cognitive aberration in schizophrenic persons—namely, a deficit in the use of contxt— may produce multiple deficits.
Conclusions pertaining to thought disorder, however, must be regarded tenta tively because of the composite nature of the thought disorder index that was used.
This measure does not take into account the fact that nonschizophrenic patients exhibit thought disorder symptoms, or the fact that schizophrenic and nonschizophrenic patients tend to display different types of thought disorder symptoms. Thus, the composite measure may not be sufficiently refined for a definitive evaluation of
Chapman’s theory. Ratings of the various types of thought disorder symptoms were made using an ordinal measurement scale, and the composite index provided a means of representing the ratings in a form that more closely approximated an interval-scaled measure. Had a more adequate index of thought disorder symptoms been available, an examination of the interrelationship among memory performance, diagnosis, and each of the thought disorder symptoms would have been possible. The fact that the present analyses indicated that the composite measure of thought disorder was related to performance suggests that further inquiry would be worthwhile. 140
Chronicitv of Schizophrenic Illness
The results indicated that, in schizophrenic subjects, chronicity was unrelated to memory performance. Thus, encoding abnormalities were not differentially present
in schizophrenic patients who differed in length of illness. These findings suggest that
the abnormalities identified here are not likely to be a product of long-term illness.
Previous studies have produced conflicting results as to whether Chapman-type
deficits are present in acute and chronic schizophrenic patients. Some researchers have
found that such deficits are observed in chronic but not acute schizophrenic patients,
whereas other researchers have observed deficits in both acute and chronic patients.
Note that the present study differs from many previous ones in the method used for
measuring chronicity. Previous researchers have often studied chronicity by dichoto
mizing patients into acute and chronic subgroups and comparing the performance of
the subgroups. In the present study, chronicity was assessed using a continuous
measurement scale, thus allowing for an examination of the relationship between
degree of chronicity and level of memory performance. The approach used here for
evaluating the effects of chronicity would appear to have greater sensitivity than the
approach used in previous studies. This factor provides added support for the assertion
that the Chapman theory deals with abnormalities related to schizophrenia, and that
such abnormalities are not attributable to long-term illness.
Concluding Remarks
The present study utilized an encoding specificity task to evaluate Chapman’s theory of schizophrenic thought disorder. Schizophrenic subjects’ performance on the recall tests was consistent with Chapman’s contention that schizophrenia is charac terized by an impaired ability to use context in responding to words. The data indicated that whereas normal subjects encoded information primarily in relation to contextual 141 cues, schizophrenic subjects encoded both context-related and context-independent information. The findings were also interpreted in terms of formulations developed in the experimental psychology literature dealing with item-specific and relational processing. Although presented with a task designed to induce relational processing of the to-be-remembered material, schizophrenic subjects appeared to encode both relational and item-specific information. Normal subjects, in contrast, appeared to process information primarily in terms of its relational aspects.
Schizophrenic subjects’ performance on the recall tasks differed from their performance on the recognition tasks. A number of explanations were considered, including the suggestion that patients’ performance on the two types of tests reflected the encoding of both context-related and context-independent information. However, further work is needed to rule out competing explanations.
An examination of the relative contribution of thought disorder and diagnosis in explaining encoding deviance suggested that diagnosis may have the greater predic tive power. Thus, although the Chapman theory was intended to explain schizophrenic thought disorder, encoding anomalies may be more related to a cognitive dysfunction associated with schizophrenia than to thought disorder per se. This suggestion is qualified by the fact that a composite measure of thought disorder was used, and thus the contribution of individual thought disorder symptoms in accounting for encoding abnormalities was not considered. The development of more refined measures for assessing thought disorder symptoms would be useful in future efforts to more clearly evaluate the relative importance of variables that are associated with anomalies in the use of contextual and relational information. A continued focus on the role of thought disorder seems particularly important because such anomalies would appear to explain speech aberrations that are prominent in schizophrenia. Symptoms such as loose associations, tangentiality, and circumstantiality all involve abnormalities in the 142 relationship between ideas, and thus may be explainable in terms of disturbances in relational processing.
The results indicated that the performance of schizophrenic subjects was not related to chronicity of illness. These findings suggest that the Chapman theory deals with a fundamental aspect of schizophrenia rather than the effects of long-term illness.
The present study is distinctive in that its evaluation of theoretical assertions concerning schizophrenic performance does not rely on the strategy of demonstrating a greater schizophrenic deficit on some tasks than on others. This study also exemplifies the advantages that can be derived from utilizing constructs and techniques developed in the field of cognitive psychology to evaluate theories of schizophrenia. The findings suggest that important advances in the understanding of schizophrenia will likely come from an active approach to theorizing along with the continued use of developments in cognitive psychology. END NOTES
1. See Chapman and Chapman (1973a) for an overview of research on the response bias theory.
2. Feinberg and Mercer (1960) did not cite any statistical techniques in evaluating their results. For the purposes of the present discussion, the investigators are assumed to have examined their data using formal statistical analyses.
3. Schizophrenia researchers typically sample subject groups that are not significantly different in terms of factors such as age and education. This practice circumvents the problems associated with evaluating data from groups that differ demographically in important ways.
4. The investigators also administered the card-sorting task under several additional conditions in order to evaluate some hypotheses that are unrelated to the present paper. The interested reader is referred to their work for further details.
5. Grimes and McGhie (1973) evaluated the data with an analysis of variance (ANOVA), and employed a .01 level of significance in order to correct for heterogeneity of variance among the subject groups. The use of an unusually stringent significance criterion seems unfounded in light of the robust nature of the ANOVA technique and the availability of nonparametric techniques such as the one used by Chapman (1958).
6. Marshall (1973), Hemsley and Hawks (1974), and Grimes and McGhie (1973) administered additional variations of the card-sorting task as a means of testing several theories other than the response bias theory. See their articles for more information.
7. Kwapil et al. (1990) devised a modified lexical decision procedure designed to take into account the potentially confounding effects of a global schizophrenic deficit on performance measures. The modified procedure involved measuring subjects’ accuracy, rather than response latency, in responding to the word pairs, and titrating subjects’ overall accuracy level through the use of degraded stimuli. See Kwapil et al.’s (1990) article for further information.
8. Guidelines for converting anti-Ach medications to a common scale of measurement are not available.
9. The terms “clue word” and “memory word” correspond to the cue word and target word, respectively, discussed earlier and in connection with Tulving’s work. The two types of words will be referred to as cues and targets in the present work, except when they are discussed in the context of instructions to subjects.
143 144
10. As applied to the measurement of interrater agreement, values of the kappa statistic represent the proportion of ratings for which two raters agree in their judgements, corrected for the proportion of chance-expected agreements. A kappa value of 0 signifies chance levels of agreement, whereas a value of 1 signifies perfect agreement Unweighted kappa provides a measure of agreement for judgements in which the two raters agree exactly. Weighted kappa provides partial credit for instances in which the raters make similar judgements.
11. The recognition tests also contained distractor items that did not include the to-be- remembered targets. As the present formulation deals with how stored information is accessed, the processes involved in detecting nontarget information are not relevant to the discussion here. APPENDIX A
SOURCES OF PATIENT SUBJECTS
145 146
Table 17 Facilities and Programs From Which Patient Subjects Were Drawn
Number of Subjects
Schizophrenic Bipolar Name and Location Group Group
Netcare Corporation 9 9 Columbus, Ohio
North Community Counseling Centers 3 0 Columbus, Ohio
Veterans Administration Outpatient Clinic 3 1 Columbus, Ohio
Northwest Counseling Services 1 0 Upper Arlington, Ohio
Concord Counseling Services 1 2 Westerville, Ohio
Harding Hospital 2 0 Worthington, Ohio
Adult Mood Disorders Clinic, Department of Psychiatry, The Ohio State University 0 4 Columbus, Ohio
Schizophrenia Research Program, Department of Psychiatry, The Ohio State University 1 1 Columbus, Ohio
Bipolar Bears 0 2 Columbus, Ohio
Recovery 0 1 Columbus, Ohio APPENDIX B
SCREENING INSTRUMENTS USED TO DETERMINE
ELIGIBILITY FOR PARTICIPATION
147 148
SCREENING WORKSHEET FOR PATIENTS
Name Date
Age______Education Marital Status S M D Sep W
Occupation______Yrs. Worked
Education and occupation of family members in subject’s present household
Have you ever been hospitalized for any reason other than psychiatric problems? ______
Have you ever had a head injury?______
Have you ever lost consciousness?______
Have you ever had problems with seizures, fits, or epilepsy?______
Have you ever had or do you now have any serious or chronic medical problems? ______
Are you presently taking any medication? (in recording, circle medications that cause drowsiness.)
Have you ever been addicted to alcohol, prescribed medication, or street drugs?
Have you ever had a drag or alcohol problem or been treated for any type of drag or alcohol problem?______
How often do drink alcohol? Frequency of use______Average amount consumed per occurrence______
What street drags have you used? (Obtain information on names of drugs used and recency, frequency, and amount of usage.) 149
SCREENING WORKSHEET FOR NORMALS
Name Date
Age_____ Education Marital Status S M D Sep W
Occupation Yrs. Worked ______
Education and occupation of family members in subject’s present household
Have you ever been hospitalized?______
Have you ever had a head injury?______
Have you ever lost consciousness?______
Have you ever had problems with seizures, fits, or epilepsy?______
Have you ever had or do you now have any serious or chronic medical problems?______
Are you presently taking any medication? (In recording, circle medications that cause (frowsiness.)
Have you ever been hospitalized for mental problems?______
Has a doctor or other professional ever given you a diagnosis involving a mental disorder?______
Have you ever seen a psychiatrist, psychologist, or other professional for mental problems?______
Have you ever been addicted to alcohol, prescribed medication, or street drugs?
Have you ever had a drag or alcohol problem or been treated for any type of drag or alcohol problem?______
How often do drink alcohol? Frequency of use______Average amount consumed per occurrence______
What street drags have you used? (Obtain information on names of drugs used and recency, frequency, and amount of usage.) APPENDIX C
WORD TRIPLETS USED IN THE CONSTRUCTION
OF MEMORY-TASK MATERIALS
150 Table 18 Practice List Triplets
Weak Cue Target
Practice List A problem help garden yard
Practice List B cake party moment wait wisdom power stomach full pass skip
Practice List C ground level whistle shout bean seed hold keep reason fact Table 19 List 1 Triplets
Weak cue Sttang-SBS Target bill sparrow bird
file law order
sorrow sugar sweet
away open close
limp finger hand
bear carpet rug thing truck car
sharp fast slow
letters question answer
rough beach sand
morning hot cold
star black white
doctor ache pain hope bulb light cloth trousers pants
ugly thin fat
refuse give take
shower song sing
hair rear end
cottage war peace Table 20 List 2 Triplets
-We.ak -C.yg______Strong cue______Target glass bath clean
mass region area
animal walk run
office table chair
lotion hard soft
plumber nickel money
boil net fish
fog look see
stem top bottom
fair stop go
club dumb stupid
mountain cube ice
quietly bed sleep
temple home house
spray rake leaves
pig fruit apple
figure knee leg
swift pilot plane
game woman man
produce labor work 154
Table 21 List 3 Triplets
Weak cue Strong cue Target cheese bad good
write thought think
race harbor boats
wish joy happy moon like love
ruler count numbers
lettuce insect bug
tie socks shoes
plum sky blue
sidewalk forest tree
talk ear hear
spring storm rain
with one two
roof dog cat
sword bow arrow
clearer noise loud
scale flute music
pound pepper salt
building crowd people
lace pelt fur Table 22 List 4 Triplets
Weak cue Strong cue Target
carry sell buy
welfare pupil student
valley dry wet
date then now
meat pony horse
candy boy girl
drums square round
silk needle thread
heavy author book
attic young old
tiger dentist teeth jump near far rock low high business town city traitor live die rest stand sit thirsty ale beer judge bible god box out in sunshine bloom flower APPENDIX D
STUDY ITEM LISTS AND MEMORY TEST ANSWER SHEETS
FOR THE MEMORY TASKS
156 157
Practice Trial 1 Study Items
problem-HELP
garden-YARD
Practice Trial 1 Answer Sheet
On the screen, you saw: problem HELP
garden YARD
Here is one type of memory test:
Garden ______
Problem ______
Here is another type of test of your memory:
Garden-Tool
Problem-Help 158
Practice Trial 2 Study Items
cake-PARTY
moment-WAIT
wisdom-POWER
stomach-FULL
pass-SKIP
Practice Trial 2 Recall Test Answer Sheet
Stomach ______
Moment ______
Pass ______
Cake ______
Wisdom ______
Practice Trial 2 Recognition Test Answer Sheet
Pass-Skip
Cake-Spoon
Wisdom-Power
Stomach-Full
Moment-Later 159
Practice Trial 3 Study Items
ground-LEVEL
whistle-SHOUT
bean-SEED
hold-KEEP
reason-FACT
Practice Trial 3 Recall Test Answer Sheet
Hold ______
Whistle ______
Bean ______
Ground ______
Reason ______
Practice Trial 3 Recognition Test Answer Sheet
Whistle-Shout
Reason-Right
Hold-Touch
Ground-Mix
Bean-Seed List Order 1
Trial 1 Study Items
Bill-BIRD
File-ORDER Sorrow-SWEET
Away-CLOSE Limp-HAND
Bear-RUG
Thing-CAR
Sharp-SLOW
Letters-ANSWER
Rough-SAND
Moming-COLD Star-WHITE
Doctor-PAIN
Hope-LIGHT
Cloth-PANTS
Ugly-FAT
Refuse-TAKE
Shower-SING
Hair-END
Cottage-PEACE List Order 1 Trial 1 Recall Test Answer Sheet
Letters
Shower
Refuse Hope
Limp
Ugly
Hair
File
Morning
Star
Bill
Sorrow
Cottage
Rough
Doctor
Thing
Bear
Cloth
Sharp
Away List Order 1
Trial 1 Recognition Test Answer Sheet
Refuse-Obey
Thing-Ball
Sharp-Slow
Hope-Idea Doctor-Pain
Star-Heaven
Cloth-Pants
Ugly-Fat
Letters-Answer
Shower-Fresh
Limp-Stick
Bear-Rug Away-Trip
File-Order
Sorrow-Sweet
Cottage-Country
Hair-Grow
Rough-Sand
Bill-Bird
Moming-Rise List Order 1
Trial 2 Study Items
Glass-CLEAN
Mass-AREA
Animal-RUN
Office-CHAIR
Lotion-SOFT
Plumber-MONEY
Boil-FISH
Fog-SEE
Stem-BOTTOM
Fair-GO
Club-STUPID
Mountain-ICE
Quietly-SLEEP
Temple-HOUSE
Spray-LEAVES
Pig-APPLE
Figure-LEG
Swift-PLANE
Game-MAN
Produce-WORK 164
List Order 1
Trial 2 Recall Test Answer Sheet
Boil ______F i g u r e ______
Swift ______
Mass ______.
Animal ______
Spray ______Temple ______
Club ______
Game ______
F o g ------
Glass ______
Stem ______-____
Produce ______
Q u i e t l y______M ountain______
Plumber ______
Lotion ______
Office ______
Pig ------Fair ______165
List Order 1
Trial 2 Recognition Test Answer Sheet
Glass-Bowl
Plumber-Money
Office-Chair
Club-Beat Animal-Run
Quietly—S till
Figure-Form
Mass-Area
Fair-Go
Spray-Air
Mountain-Ice
Game-Hurt
Temple-Cave
Stem-Base
Lotion-Soft Swift-Deer
Pig-Apple
Fog-See
Produce-Cause
Boil-Fish 166
List Order 1
Trial 3 Study Items
Cheese-GOOD
Write-THINK
Race-BOATS
Wish-HAPPY Moon-LOVE
Ruler-NUMBERS
Lettuce-BUG
Tie-SHOES
Plum-BLUE
Sidewalk-TREE
Talk-HEAR
Spring-RAIN
With-TWO
Roof-CAT
Sword-ARROW
Clearer-LOUD
Scale-MUSIC
Pound-SALT
B uilding-PEOPLE
Lace-FUR List Order 1
Trial 3 Recall Test Answer Sheet
One Dog Sky Count Bow Joy Bad Thought Crowd Storm Harbor Pelt Flute Pepper Insect Forest Like Socks Noise List Order 1
Trial 3 Recognition Test Answer Sheet
Harbor-Boats Cook-Food Joy-Happy Noise-Loud
Web-Spider Ear-Hear
Storm-Rain Bow-Arrow
Night-Day Second-Minute
One-Two Insect-Bug
Blow-Wind Bad-Good
Count-Numbers Hammer-Nail
Husband-Wife Ill-Sick
Over-Under Dog-Cat
Pepper-Salt Sky-Blue
Lead-Follow Chimney-Smoke
Scissors-Cut Thought-Think
Flute-Music Rifle-Gun
Like-Love Down-Up
Highway-Road Queen-King
Robber-Thief Crowd-People
Off-On Entrance-Exit
Forest-Tree Pelt-Fur
Anger-Mad Socks-Shoes List Order 1
Trial 4 Study Items
BUY
STUDENT
WET
NOW HORSE
GIRL
ROUND
THREAD
BOOK
OLD
TEETH
FAR HIGH
CITY
DIE
SIT
BEER
GOD
IN
FLOWER List Order 1
Trial 4 Recall Test Answer Sheet
Dry
Ale
Out Needle
Pony
Boy
Square
Bloom
Dentist
Pupil
Sell Low
Author
Near
Stand
Town
Live
Then
Bible
Young List Order 1
Trial 4 Recognition Test Answer Sheet
Here-There Shallow-Deep
Square-Round Always-Never
Army-Navy Low-High
Young-Old Butter-Bread
Sell-Buy Pupil-Student
Me-You Bible-God
Needle-Thread Hide-Seek
Long-Short Dentist-Teeth
Out-In Friend-Foe
Near-Far Punch-Hit
Broader-Wider Bloom-Flower
Circus-Clown Pony-Horse
Author-Book Ale-Beer
Stand-Sit Grass-Green
Tin-Can Ocean-Water
Small-Large Boy-Girl
Live-Die Lost-Found
Infant-Baby Gown-Dress
Town-City Cow-Milk List Order 2
Trial 1 Study Items
Glass-CLEAN
Mass-AREA
Animal-RUN
Office-CHAIR
Lotion-SOFT
Plumber-MONEY
Boil-FISH
Fog-SEE
Stem-BOTTOM
Fair-GO Club-STUPID
Mountain-ICE
Quietly-SLEEP
Temple-HOUSE
Spray-LEAVES
Pig-APPLE
Figure-LEG
Swift-PLANE
Game-MAN
Produce-WORK 173
List Order 2
Trial 1 Recall Test Answer Sheet
Boil ______
F i g u r e ______
Swift ______
Mass ______
Animal ______
Spray ------Temple _____ ------
Club ______
Game ______
Fog ------—
Glass ,______
S t e m ______Produce ______
Quietly ______Mountain______
Plumber ______
Lotion ______
Office ______-
Pig Fair 174
List Order 2 Trial 1 Recognition Test Answer Sheet
Glass-Bowl
Plumber-Money
Office-Chair
Club-Beat
Animal-Run
Quietly—Still
Figure-Form
Mass-Area
Fair-Go
Spray-Air
Mountain-Ice
Game-Hurt
Temple-Cave
Stem-Base
Lotion-Soft
Swift-Deer
Pig-Apple
Fog-See
Produce-Cause
Boil-Fish 175
List Order 2
Trial 2 Study Items
Cheese-GOOD
Write-THINK
Race-BOATS Wish-HAPPY
Moon-LOVE
Ruler-NUMBERS
Lettuce-BUG
Tie-SHOES
Plum-BLUE
Sidewalk-TREE
Talk-HEAR
Spring-RAIN
With-TWO
Roof-CAT
Sword-ARROW Clearer-LOUD
Scale-MUSIC
Pound-SALT
Building-PEOPLE
Lace-FUR List Order 2 Trial 2'Recall Test Answer Sheet
With ______
Roof ______
Plum ______
Ruler ______
Sword ______
Wish ______
Cheese ______■
Write ______
B u ild in g ______
Spring ------
Race ______
Lace ______
Scale ______
Pound ______
Lettuce ______Sidewalk ______
Moon ______
Tie ______
Clearer List Order 2
Trial 2 Recognition Test Answer Sheet
Race-Boats
Wish-Luck
Spring-Rain
With-Alone
Sword-Brave
Pound-Salt
Scale-Notes
Moon-Love
Sidewalk-Path
Clearer-Loud
Talk-Hear
Cheese-Smell
Lettuce-Bug
Write-Draw
Roof-Fall
Plum-Blue
Ruler-Numbers
Building-Wood
Lace-Pretty
Tie-Shoes List Order 2
Trial 3 Study Items
Carry-BUY
Welfare-STUDENT
Valley-WET
Date-NOW
Meat-HORSE
Candy-GIRL
Drums-ROUND
Silk-THREAD
Heavy-BOOK
Attic-OLD
Tiger-TEETH
Jump-FAR
Rock-HIGH
Business-CITY
Traitor-DIE
Rest—SIT
Thirsty-BEER
Judge-GOD Box-IN
Sunshine-FLOWER List Order 2
Trial 3 Recall Test Answer Sheet
Dry
Ale
Out Needle
Pony
Boy
Square
Bloom
Dentist
Pupil
Sell
Low
Author
Near
Stand
Town
Live
Then
Bible
Young 180
List Order 2 Trial 3 Recognition Test Answer Sheet
Square-Round Cook-Food
Young-Old Town-City
Web-Spider Dry-Wet
Sell-Buy Ale-Beer
Night-Day Second-Minute
Out-In Pupil-Student
Blow-Wind Bible-God Near-Far Hammer-Nail
Husband-Wife Ill-Sick
Over-Under Dentist-Teeth
Needle-Thread Bloom-Flower
Lead-Follow Chimney-Smoke
Scissors-Cut Pony-Horse Author-Book Rifle-Gun
Stand-Sit Down-Up
Highway-Road Queen-King
Robber-Thief Boy-Girl
Off-On Entrance-Exit
Live-Die Low-High
Anger-Mad Then-Now List Order 2
Trial 4 Study Items
BIRD
ORDER
SWEET
CLOSE
HAND
RUG
CAR
SLOW
ANSWER SAND
COLD
WHITE
PAIN
LIGHT
PANTS
FAT
TAKE
SING
END
PEACE 182
List Order 2
Trial 4 Recall Test Answer Sheet
Question ______
Song ------
Give ______
Bulb ______F i n g e r ______
Thin ______
R e a r ______
Law
Hot ______
Black ______
Sparrow ------'
Sugar ______
War ______
B e a c h ______.
Ache ______-
Truck ______
Carpet ______
T rousers______
F a s t ______
O p e n ______183
List Order 2
Trial 4 Recognition Test Answer Sheet
Here-There Shallow-Deep
Black-White Always-Never
Army-Navy Carpet-Rug
Truck-Car Butter-Bread
Fast-Slow Open-Close
Me-Y ou Law-Order
Bulb-Light Hide-Seek
Long-Short Trousers-Pants Ache-Pain Friend-Foe Give-Take Punch-Hit
Broader-Wider War-Peace
Circus-Clown Rear-End
Thin-Fat Beach-Sand
Sugar-Sweet Grass-Green
Tin-Can Ocean-Water
Small-Large Sparrow-Bird
Question-Answer Lost-Found
Infant-Baby Gown-Dress
Song-Sing Cow-Milk
Finger-Hand Hot-Cold 184
List Order 3
Trial 1 Study Items
Cheese-GOOD Write-THINK
Race-BOATS
Wish-HAPPY
Moon-LOVE
Ruler-NUMBERS
Lettuce-BUG
Tie-SHOES Plum-BLUE
Sidewalk-TREE
Talk-HEAR
Spring-RAIN
With-TWO
Roof-CAT
Sword-ARROW
Clearer-LOUD
Scale-MUSIC
Pound-SALT
Building-PEOPLE
Lace-FUR 185
List Order 3
Trial 1 Recall Test Answer Sheet
With ______
R o o f ______
Plum ______Ruler ------
Sword ______
W i s h ______—
Cheese ______
Write -______
Building ______.___
Spring ------
Race ------
Lace ______
Scale ______
Pound ______
Lettuce ______Sidewalk ______
Moon ______
Tie ______Clearer ______
Talk _____ - List Order 3
Trial 1 Recognition Test Answer Sheet
Race-Boats
Wish-Luck
Spring-Rain
With-Alone
Sword-Brave
Pound-Salt
Scale-Notes
Moon-Love
Sidewalk-Path
Clearer-Loud
Talk-Hear
Cheese-Smell
Lettuce-Bug
Write-Draw
Roof-Fall
Plum-Blue
Ruler-Numbers
Building-Wood
Lace-Pretty
Tie-Shoes List Order 3 Trial 2 Study Items
Carry-BUY
Welfare-STUDENT
Valley-WET
Date-NOW Meat-HORSE
Candy-GIRL
Drums-ROUND
Silk-THREAD
Heavy-BOOK
Attic-OLD
Tiger-TEETH
Jump-FAR
Rock-HIGH
Business-CITY
Traitor-DIE
Rest-SIT
Thirsty-BEER
Judge-GOD
Box-IN
Sunshine-FLOWER 188
List Order 3 Trial 2 Recall Test Answer Sheet
V a l l e y______Thirsty ______
B o x ______
Silk ______
M e a t ______Candy ______Drums ______
Sunshine______
Tiger ------
Welfare ______
C a r r y______
Rock ______
Heavy ______
Jump ______
Rest ______Business______
T r a i t o r ______
Date ______
Judge ______
A t t i c ______189
List Order 3 Trial 2 Recognition Test Answer Sheet
Drums-Play
Attic-Dust
Carry-Buy
Silk-Material
Box-In
Jump-Far
Valley-Below
Rest-Sit
Traitor-Hang
Business-City
Heavy-Lift
Thirsty-Beer
Welfare-Student
Judge-Honest
Tiger-Teeth
Sunshine-Weather
Meat-Market
Candy-Taste
Rock-High
Date-Now List Order 3
Trial 3 Study Items
Bill-BIRD
File-ORDER
Sorrow-SWEET
Away-CLOSE Limp-HAND
Bear-RUG
Thing-CAR
Sharp-SLOW
Letters-ANSWER
Rough-SAND
Moming-COLD
Star-WHITE
Doctor-PAIN
Hope-LIGHT
Cloth-PANTS
Ugly-FAT
Refuse-TAKE
Shower-SING
Hair-END
Cottage-PEACE 191
List Order 3
Trial 3 Recall Test Answer Sheet
Question ______
Song ______
Give ______.
Bulb ______Finger ______
Thin ______
Rear ______Law ______
Hot ______
Black ______
Sparrow ______
Sugar ______
War ______
Beach ______
Ache ______Truck ______
Carpet ______
Trousers ______
Fast ______
Open ______List Order 3
Trial 3 Recognition Test Answer Sheet
Give-Take Cook-Food
Truck-Car Song-Sing
Web-Spider Finger-Hand
Bulb-Light Carpet-Rug
Night-Day Second-Minute
Fast-Slow Open-Close
Blow-Wind Sugar-Sweet
Ache-Pain Hammer-Nail
Husband-Wife Dl-Sick
Over-Under Rear-End
Trousers-Pants War-Peace
Lead-Follow Chimney-Smoke
Scissors-Cut Law-Order
Black-White Rifle-Gun
Thin-Fat Down-Up
Highway-Road Queen-King
Robber-Thief Beach-Sand
Off-On Entrance-Exit
Question-Answer Sparrow-Bird
Anger-Mad Hot-Cold List Order 3
Trial 4 Study Items
CLEAN
AREA
RUN
CHAIR
SOFT
MONEY
FISH
SEE
BOTTOM
GO
STUPID
ICE
SLEEP
HOUSE
LEAVES
APPLE
LEG
PLANE
MAN
WORK 194
List Order 3
Trial 4 Recall Test Answer Sheet
N e t ______K n e e ______
Pilot ______Region______
Walk ______
Rake ______
Home ______
Dumb ______
Woman______
L o o k ______
Bath ______
Top ______
Labor ______
Bed ______
C u b e ______
N ick el ______
H a r d ______
T a b l e ______
Fruit ______
Stop ______195
List Order 3
Trial 4 Recognition Test Answer Sheet
Here-There Shallow-Deep
Bath-Clean Always-Never
Army-Navy Woman-Man
Nickel-Money Butter-Bread
Labor-Work Home-House Me-You Top-Bottom
Table-Chair Hide-Seek
Long-Short Stop-Go
Walk-Run Friend-Foe
Dumb-Stupid Punch-Hit
Broader-Wider Hard-Soft
Circus-Clown Fruit-Apple
Knee-Leg Look-See
Region-Area Grass-Green Tin-Can Ocean-Water
Small-Large Pilot-Plane
Bed-Sleep Lost-Found
Infant-Baby Gown-Dress
Rake-Leaves Cow-Milk
Cube-Ice Net-Fish List Order 4
Trial 1 Study Items
Carry-BUY Welfare-STUDENT
Valley-WET
Date-NOW
Meat-HORSE
Candy-GIRL
Drums-ROUND
Silk-THREAD
Heavy-BOOK
Attic-OLD
Tiger-TEETH
Jump-FAR
Rock-HIGH
Business-CITY
Traitor-DIE
Rest-SIT Thirsty-BEER
Judge-GOD
Box-IN
Sunshine-FLOWER 197
List Order 4
Trial 1 Recall Test Answer Sheet
Valley ______
Thirsty ______
Box ______
Silk ______
Meat ______
Candy ______
Drums ______
Sunshine______
Tiger ------
Welfare ______
Carry ______
Rock ______
Heavy ______
Jump ______
Rest ______
Business ______
Traitor ______
Date ______
Judge ______
Attic ______198
List Order 4
Trial 1 Recognition Test Answer Sheet
Drums-Play
Attic-Dust
Carry-Buy
Silk-Material
Box-In
Jump-Far
Valley-Below
Rest-Sit Traitor-Hang
Business-City
Heavy-Lift
Thirsty-Beer
Welfare-Student
Judge-Honest
Tiger-Teeth
Sunshine-Weather
Meat-Market
Candy-Taste
Rock-High
Date-Now List Order 4 Trial 2 Study Items
Bill-BIRD
File-ORDER
Sorrow-SWEET
Away-CLOSE
Limp-HAND
Bear-RUG
Thing-CAR
Sharp-SLOW
Letters-ANSWER
Rough-SAND
Moming-COLD
Star-WHITE
Doctor-PAIN
Hope-LIGHT
Cloth-PANTS Ugly-FAT
Refiise-TAKE
Shower-SING
Hair-END
Cottage-PEACE List Order 4
Trial 2 Recall Test Answer Sheet
Letters Shower Refuse Hope Limp Ugly Hair File Morning Star Bill Sorrow Cottage Rough Doctor Thing Bear Cloth Sharp Away 201
List Order 4 Trial 2 Recognition Test Answer Sheet
Refuse-Obey
Thing-Ball
Sharp-Slow
Hope-Idea Doctor-Pain
Star-Heaven
Cloth-Pants
Ugly-Fat
Letters-Answer
Shower-Fresh
Limp-Stick
Bear-Rug
Away-Trip
File-Order
Sorrow-Sweet
Cottage-Country
Hair-Grow
Rough-Sand
Bill-Bird
Morning-Rise List Order 4
Trial 3 Study Items
Glass-CLEAN
Mass-AREA
Animal-RUN
Office-CHAIR
Lotion-SOFT
Plumber-MONEY
Boil-FISH Fog-SEE
Stem-BOTTOM Fair-GO Club-STUPID
Mountain-ICE
Quietly-SLEEP
Temple-HOUSE
Spray-LEAVES
Pig-APPLE
Figure-LEG Swift-PLANE
Game-MAN
Produce-WORK 203
List Order 4 Trial 3 Recall Test Answer Sheet
Net ______Knee ______
Pilot ______Region ______
Walk ______
Rake ______
Home ______;—
Dumb ______
W om an______Look ______
Bath ______
Top ______Labor ______
Bed ______
Cube ______
Nickel ______
Hard ______
Table ______
Fruit ______
Stop ______204
List Order 4
Trial 3 Recognition Test Answer Sheet
Bath-Clean Cook-Food
Nickel-Money Rake-Leaves
Web-Spider Cube-Ice
Table-Chair Hard-Soft
Night-Day Second-Minute
Dumb-Stupid Home-House
Blow-Wind Top-Bottom
Walk-Run Hammer-Nail
Husband-Wife Ill-Sick
Over-Under Woman-Man
Bed-Sleep Pilot-Plane
Lead-Follow Chimney-Smoke
Scissors-Cut Fruit-Apple
Knee-Leg Rifle-Gun
Region-Area Down-Up
Highway-Road Queen-King
Robber-Thief Look-See
Off-On Entrance-Exit
Stop-Go Labor-Work
Anger-Mad Net-Fish 205
List Order 4
Trial 4 Study Items
GOOD
THINK
BOATS
HAPPY
LOVE
NUMBERS
BUG
SHOES
BLUE
TREE
HEAR
RAIN
TWO
CAT
ARROW
LOUD
MUSIC
SALT
PEOPLE
FUR 206
List Order 4
Trial 4 Recall Test Answer Sheet
One ______
Dog ______
Sky ______
Count ______
Bow ______
Joy ______
Bad ______
Thought______
Crowd ______
Storm ______
Harbor ______
Pelt ______
Flute ______
Pepper ______
Insect ______
Forest ______
Like ______
Socks ______
Noise ______
Ear List Order 4
Trial 4 Recognition Test Answer Sheet
Here-There Shallow-Deep
Insect-Bug Always-Never
Army-Navy Pelt-Fur
Joy-Happy Butter-Bread
Storm-Rain Crowd-People
Me-You Thought-Think
One-Two Hide-Seek
Long-Short Dog-Cat
Bad-Good Friend-Foe
Pepper-Salt Punch-Hit
Broader-Wider Sky-Blue «Circus-Clown Count-Numbers
Flute-Music Bow-Arrow
Like-Love Grass-Green
Tin-Can Ocean-Water
Small-Large Harbor-Boats
Forest-Tree Lost-Found
Infant-Baby Gown-Dress
Noise-Loud Cow-Milk
Ear-Hear Socks-Shoes APPENDIX E
SAMPLE CONSENT FORM
208 209
CONSENT TO INVESTIGATIONAL TREATMENT OR PROCEDURE
I,______, hereby authorize or direct Herbert Mirels, Ph.D or associates or assistants of his or her choosing, to perform the following treatment or procedure upon myself: To administer tests as described below.
The experimental (research) portion of the treatment or procedure is: A vocabulary test, a test for memory of numbers, and a series of memory tests involving words. I will also be asked to talk about some common, everyday topics.
This is done as part of an investigation entitled: Thinking and Remembering.
1. Purpose of the procedure or treatment: To study the relationship between thinking and memory and to further understand how people differ in thinking and remembering.
2. Possible appropriate alternative methods of treatment: Not to participate.
3. Discomforts and risks reasonably to be expected: There may be some minor frustration during the memory tests in attempting to recall the words. The amount of frustration that is expected is similar to the discomfort that results, in everyday life, in trying to recall a memory that has been lost or forgotten. In participating in this study, if the memory tests cause more than minor frustration, in the opinion of either the participant or the experimenter, the study will be terminated immediately.
4. Possible benefits for subjects/society: I will receive $25 upon completion of my participation. My participation will also allow scientists to develop a better under standing of memory and thought processes in people.
5. Anticipated duration of subject’s participation: One and one-half hours.
I hereby acknowledge that Herbert Mirels, Ph.D has provided information about the procedure described above, about my rights as a subject, and he/she answered all questions to my satisfaction. I understand that I may contact him/her should I have additional questions. He/She has explained the risks described above and I understand them; he/she has also offered to explain all possible risks or complications.
I understand that, where appropriate, the U.S. Food and Drug Administration may inspect records pertaining to this study. I understand further that records obtained during my participation in this study may be made available to the sponsor of this study and that the records will not contain my name or other personal identifiers. Beyond this, I understand that my participation will remain confidential.
I understand that I am free to withdraw my consent and participation in this project at any time after notifying the project director without prejudicing future care. No guarantee has been given to me concerning this treatment or procedure. 210
In the unlikely event of injury resulting from participation in this study, I understand that immediate medical treatment is available at University Hospital of The Ohio State University. I also understand that the costs of such treatment will be at my expense and that financial compensation is not available. Questions about this should be directed to the Human Subjects Review Office at 292-9046.
I have read and fully understand the consent form. I sign it freely and voluntarily. A copy has been given to me.
Date______Time______AM PM
Signed ______(Subject)
(Person Authorized to Consent for Subject— If Required)
(Witness — If Required)
(Witness — If Required)
I certify that I have personally completed all blanks in this form and explained them to the subject or his/her representative before requesting the subject or his/her representative to sign it.
Signed ______(Signature of Project Director or his/her Authorized Representative) APPENDIX F
MEMORY TASK INSTRUCTIONS
211 212
Introductory Instructions
The last part of the testing today involves some memory tasks. I will be showing you sets of slides containing words on the screen in front of you. After each set of slides is shown, your memory for the words will be tested. First, we will complete three brief practice trials so you can become familiar with the testing procedure. Then, I will give you four longer trials.
[Display the first slide from the first practice set (PROBLEM-HELP). Point to the cue and target words on the displayed slide when they are mentioned in the instructions.] Here is the first slide from the first practice set. As you can see, there are two words here, one above the other. The word on top is in small letters and the word on the bottom is in large or capital letters. I am going to call the bottom word here the “memory word” because this is the word that I am going to ask you to remember. Now notice that the top word and the bottom word, the memory word, are related to one another in some way in terms of meaning. Do you see how the words are related? I am going to call this top word, in small letters, the “clue word” because it is related to the memory word. All of the other slides will be like this one. The other slides will have a clue word, in small letters, on top, and a memory word, in capital letters, on the bottom. And the clue and memory words will be related to one another in terms of their meaning.
[Continue pointing to the cue and target words whenever they are mentioned in the following instructions.] When I show you a set of slides, the slides will appear one by one for 5 seconds apiece. Your main task, as the slides are shown, is to study and try to remember as many of the memory words as possible. In addition, pay attention to the clue word on each slide, and note in your mind the relationship between the memory word and the clue word. Doing so will help you in the memory tests; you will see in a minute, when I give you the tests, what I mean. Remember, though, your main job is to study and remember as many of the memory words as possible.
Practice Trial 1
[Set up the slide projector so that it is ready to show the first set of practice slides. The projector lamp should be left on, and the screen should be blank.] Now I will show you the first set of practice slides. Whenever I show you a set of slides, the screen will be blank, like it is now, and I will tell you when the slides are about to appear. Each slides will appear for 5 seconds. After the last slide is shown, the screen will be blank again. Then I will give you the memory tests. The first set of slides will appear in a moment. [Show the first set of practice slides.]
[Give the subject the memory-test answer sheet corresponding to the first set of practice slides. In contrast to the other memory-test answer sheets, the answer sheet for the first practice trial contains the word pairs from the slides, listed at the top, above the recall and recognition tests. Point to the various parts of the answer sheet as they are discussed.] Each time you see a set of slides, I will give you two different memory tests. This time, to help illustrate how the tests are to be completed, the words that you saw on the screen are printed at the top of the answer sheet; also, both of the memory tests are here on the same answer sheet. After this time, though, the words from the 213
slides will not appear on the answer sheet, and each of the tests will be given on a separate sheet of paper.
[Continue to point to the various parts of the answer sheet when they are mentioned in the instructions.] Here is the first memory test. For the first test, you will be given a list of the clue words that appeared on the screen. See, PROBLEM and GARDEN were the clue words on the slides, and they are listed down here in the memory test. For this test, you are to write down next to each clue word the memory word that goes with it. The first clue word listed is GARDEN. What would you write in the blank next to it? [Review instructions, if necessary, until the subject produces the correct answer.] Go ahead and write it down. [Wait while subject writes.] Now how about the second one- what goes in the blank for that one? [Subject responds.] Go ahead and write down that one too. Good.
[Continue to point to the various parts of the answer sheet as they are discussed.] Now look down here, at the second memory test. For this one, you will be given pairs of words. In each pair, the word on the left will always be a clue word from the screen. See, PROBLEM and GARDEN are the left-hand members of the pairs, and see, [point to the word pairs at the top of the answer sheet] they were presented as clue words on the slides. Now, the right-hand member of each pair will be either the memory word that goes with the clue word or it will be a completely new word that was not on the screen. For this memory test, you are to look at each pair and either circle or cross out the word on the right side. Circle the word if it is a memory word that you saw on the screen, and cross it out if it is not a memory word from the screen. The first pair in this test is GARDEN-TOOL. Would you circle or cross out the word TOOL? [Review the instructions if needed.] Go ahead and do that. The second pair is PROBLEM-HELP. What would you do for that one? [Subject responds.] That’s right. Go ahead and do that.
One thing to notice about the memory tests is that the order in which the words appear on the answer sheet is different from the order in which they appear on the screen. See, [point to the top of the answer sheet], you saw PROBLEM-HELP first and GARDEN- TOOL second on the screen, but see, [point to the memory-test items] you were tested for your memory of GARDEN-TOOL first and PROBLEM-HELP second. The other trials will be similar to this one in that you will see the words in one order, but will be tested in a different order. [Remove the answer sheet.]
Now you have an idea of how the memory tests are to be completed. There are two more practice trials so you will have a chance to become more familiar with the procedure. All of the other trials will proceed in the same manner as that of the trial we just completed, except, from now on, I am going to ask you to carry out one more task. From now on, each time you see a set of slides, I will have you work on what we call a “number task” right after you see the slides but before you receive the memory tests. You will need this sheet of paper here to do the number tasks. [Hand the subject the answer sheet for the number tasks.] From now on, after I show you a set of slides, I will be asking you to either circle or cross out either the even or the odd numbers. For example, I might say “Please circle the even numbers,” or, I might say “Please cross out the odd numbers.” The exact instructions will differ from trial to trial. When you hear the instructions, go ahead and begin working. Start here [point to the first number on the answer sheet] with the first number and work across the first row of numbers before you move down and work on the next row. Work as quickly as you can without skipping any. I will have you work on the task for 30 seconds. At the end of 30 214
seconds, I will say “stop” and then give you the memory tests. Each time you begin a new number task, start with the next section of numbers on the answer sheet. [Point to each section of numbers.]
So, on each trial, first you will see the slides, then you will work on the number task, and then you will take the memory tests. Don’t worry about forgetting what you are supposed to do at each step, because I let you know.
Practice Trial 2
Now here is the second set of practice slides. This set has more slides in it than the last one. This time, there are five slides. The first slide in this set will appear in a moment. [Show the second set of practice slides.] For the number task, please circle the even numbers. Circle the even numbers. [Allow the subject to work for 30 seconds.] Stop. [Give the subject the answer sheet for the recall test.] Here is the first memory test for the words you just saw. Do you remember how to complete the test? [Review instructions if necessary.] Let me know when you have finished. [Allow the subject to work at his or her own pace. Most subjects will complete the test within a minute. If a subject has not finished within a minute and a half, ask how he or she is doing and say: “Don’t worry too much if you can’t remember all of the words. Most people find that if they can’t recall a word after a short time, spending more time doesn’t really help much.” In any case, wait until the subject indicates he or she has finished before proceeding.] Sometimes people ask me whether, if they are not sure of an answer, they should guess. I tell them that if they think they might know an answer, they should go ahead and write it down—even if they are not entirely sure. However, don’t write down anything or make wild guesses. [Remove the recall-test answer sheet and present the recognition-test answer sheet.] Here is the second memory test. Do you remember what to do for this one? [Review instructions if necessary.] Let me know when you have finished. [If a subject is still working on the test after a minute and a half elapses, say: “Don’t worry too much if you’re not sure of all of your answers. Usually, your first hunch is the best thing to follow.” When the subject returns the answer sheet, check it to ensure he or she responded to every item. If any items are unmarked, return the answer sheet and ask the subject to complete them.] One way that this memory test is different from the first one is that for this one, you should write something down for each item, and you should guess if you are not sure. Before we go on, do you have any questions?
Practice Trial 3
Now here is the next set of practice slides. This set has five slides in it. The first slide will appear in a moment. [Show the third set of practice slides.] For the number task, remember to begin working on the next section of numbers on your answer sheet. Please cross out the odd numbers. Cross out the odd numbers. [Allow the subject to work for 30 seconds.] Stop. [Give the subject the answer sheet for the recall test.] Here is the first memory test. Let me know when you have finished. [For this and all other memory tests, allow the subject to work at his or her own pace, but, if necessary, discourage the subject from lingering over his or her responses for an undue length of time. When the subject has finished, remove the recall-test answer sheet and replace it with the recognition-test answer sheet.] Here is the second test. Let me know when you have finished. [For this and all other recognition tests, check the answer sheet to 215 ensure that the subject has marked every item; ask the subject to complete any unmarked items.]
Experimental Trial 1
Now that we have completed the practice trials, we will move on to the main trials. The main trials will proceed in the same manner as that of the practice trials, except that now, there will be 20 slides in each set. Let’s go ahead and get started. The first slide in this set will appear in a moment. [Show the first set of the main slides.] For the number task, please cross out the even numbers. Cross out the even numbers. [Allow the subject to work on the number task for 30 seconds.] Stop. Here is the first memory test. [Give the subject the recall-test answer sheet. Most subjects will be able to complete the experimental recall tests in 5 to 7 minutes. When the subject has finished the test, remove the answer sheet and pass out the recognition-test answer sheet.] Here is the second memory test. [Most subjects will complete the recognition tests of the experimental trials in 5 to 7 minutes. Collect the answer sheet when the subject has finished.]
Experimental Trial 2
Here is the next set of slides. The first slide will appear in a moment. [Show the second set of the main slides.] For the number task, please circle the odd numbers. Circle the odd numbers. [Wait 30 seconds.] Stop. Here is the first memory test. [Give the subject the recall-test answer sheet. After the subject has finished, remove the answer sheet and present the recognition-test answer sheet.] Here is the second memory test. [Collect the answer sheet when the subject has finished.]
Experimental Trial 3
Here is the next set of slides. The first slide will appear in a moment. [Show the third set of the main slides.] For the number task, please cross out the even numbers. Cross out the even numbers. [Wait 30 seconds.] Stop. Here is the first memory test. [Pass out the recall-test answer sheet.] This memory test is different from the ones you had before. This test does not contain the clue words that were on the screen. Instead, you have been given new clue words. Each of the words listed here [point to the words on the answer sheet] is related in meaning to one of the memory words from the screen. For this test, look at each of the new clue words and, for as many as you can, write down the memory word that goes with each one. [When the subject has finished, remove the answer sheet and pass out the recognition-test answer sheet. Point to the various parts of the answer sheet as they are explained.] This memory test is also different from the ones you had before. For this test, you have been given pairs of words, like before, but now there are more pairs than before. Also, before, the left- hand side of each pair was a clue word from the screen; in this test, the left-hand words are new words that did not appear on the screen. You’ll notice that in each of the pairs in this test, the word on the left and the word on the right are related to one another in some way in terms of meaning. In some of the pairs, the word on the right is a memory word from the screen. In other pairs, the right-hand word is a new word that did not appear on the screen. For this test, you are to look at each pair and mark the word on the right. Circle the word on the right if it is a memory word from the screen, and cross 216 it out if it is a new word that did not appear on the screen. Remember, the words on the left are all new, and you are to decide whether the words on the right are memory words or new words. [Collect the answer sheet when the subject has finished.]
Now here is the last set of slides. For this set, each slide will contain only one word. There will be no clue words on the slides. All of the words that you will see are memory words, and you should try to remember them all. Later, when I give you the memory tests, you will receive clue words to help you remember the memory words. The first slide will appear in a moment [Show the fourth set of slides.] For the number task, please circle the odd numbers. Circle the odd numbers. [Wait 30 seconds.] Stop. [Pass out the recall-test answer sheet.] Each of the words listed here [point to the words on the answer sheet] is related in meaning to one of the words from the screen. Write down the memory word that goes with each one of the words listed here. [When the subject has finished with the recall test, remove the answer sheet and present the recognition-test answer sheet. Point to the various parts of the answer sheet as they are discussed.] Here are pairs of words in which the word on the left and the word on the right are related to one another. In some of the pairs, the word on the right is a memory word from the screen. In other pairs, the right-hand word is a new word that did not appear on the screen. You are to look at each pair, and circle or cross out the word on the right. Circle the word on the right if it is a memory word from the screen, and cross it out if it is a new word that did not appear on the screen. [Collect the answer sheet when the subject has finished.] APPENDIX G
LIST ORDERS OF THE MEMORY TASKS
217 Table 23
Latin Square Used in the Derivation of the Four List Orders
Trial
List Order 1 2 3 4
1 1 2 3 4
2 2 3 4 1
3 3 4 1 2
4 4 1 2 3
Note. Each entry denotes the word list, numbered according to Appendices C and D, that was used to develop the study and test items. REFERENCES
American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (rev. 3rd ed.). Washington, DC: Author.
Andreasen, N. C. (1979a). Thought, language, and communication disorders. I. Clinical assessment, definition of terms, and evaluation of their reliability. Archives of General Psychiatry. 36.1315-1321.
Andreasen, N. C. (1979b). Thought, language, and communication disorders, n. Diagnostic significance. Archives of General Psychiatry. 26,1325-1330.
Andreasen, N. C. (1986). Scale for the Assessment of Thought, Language, and Communication (TLC). Schizophrenia Bulletin. 12,473-482.
Andreasen, N. C., & Grove, W. M. (1986). Thought, language, and communication in schizophrenia: Diagnosis and prognosis. Schizophrenia Bulletin. 12,348-359.
Bellezza, F. S., Cheesman, F. L., n, & Reddy, B.G. (1977).Organization and semantic elaboration in free recall. Journal of Experimental Psychology: Human Leamingand Memory. 2,539-550.
Benjamin, T. B., & Watt, N. F. (1969). Psychopathology and semantic interpretation of ambiguous words. Journal of Abnormal Psychology. 74.706-714.
Berenbaum, H., Oltmanns, T. F., & Gottesman 1.1. (1985). Formal thought disorder in schizophrenics and their twins. Journal of Abnormal Psychology. 9.4.3-16.
Bilodeau, E. A., & Howell, D. C. (1965). Free association norms bv discrete and continued methods (Contract Nonr-475(10)). Washington, DC: Office of Naval Research.
Blaney, P. H. (1974). Two studies on the language behavior of schizophrenics. Journal ofAbnormal Psychology. 22,23-31. Bleuler, E. (1950). Dementia praecox or the group of schizophrenias (J. Zinkin, Trans.). New York: International Universities Press. (Original work published 1911)
Boland, T. B., & Chapman, L. J. (1971). Conflicting predictions from Broen’s and Chapman’s theories of schizophrenic thought disorder. Journal of Abnormal Psychology. 2&, 52-58.
Bolles, M., & Goldstein, K. (1938). A study of the impairment of “abstract behavior” in schizophrenic patients. Psychiatric Quarterly. 12, 42-65.
219 220
Braff, D. L., & Saccuzzo, D. P. (1985). The time course of information-processing deficits in schizophrenia. American Journal of Psychiatry. 142. 170-174.
Broen, W. E., Jr. (1966). Response disorganization and breadth of observation in schizophrenia. Psychological Review. 22, 579-585. Broen, W. E., Jr., & Storms, L. H. (1961). A reaction potential ceiling and response decrements in complex situations. Psychological Review. 68.405-415.
Callahan, L. A., & Saccuzzo, D. P. (1986). Associative intrusions in the verbal behavior of the first-degree relatives of adult schizophrenics. A preliminary study. Journal of Nervous and Mental Disease. 124,240-242.
Cermak, L. S., & Stiassny, D. (1982). Recall failure following successful generation and recognition of responses by alcoholic Korsakoff patients. Brain and Cognition. 1.165-176.
Cermak, L. S., Uhly, B., & Reale, L. (1980). Encoding specificity in the alcoholic Korsakoff patient. Brain and Language. H , 119-127.
Chapman, L. J. (1956a). Distractibility in the conceptual performance of schizophrenics. Journal of Abnormal and Social Psychology. 52,286-291.
Chapman, L. J. (1956b). The role of type of distracter in the “concrete” conceptual performance of schizophrenics. Journal of Personality.25, 130-141. Chapman, L. J. (1958). Intrusion of associative responses into schizophrenic conceptual performance. Journal of Abnormal and Social Psychology. 56,374- 379.
Chapman, L. J. (1960). Confusion of figurative and literal usages of words by schizophrenics and brain damaged patients. Journal of Abnormal and Social Psychology. £Q, 412-416.
Chapman, L. J. (1961). A reinterpretation of some pathological disturbances in conceptual breadth. Journal of Abnormal and Social Psychology.6 2514-519. ,
Chapman, L. J., & Chapman, J. P. (1965). The interpretation of words in schizophrenia. Journal of Personality and Social Psychology. 1 , 135-146.
Chapman, L. J., & Chapman, J. P. (1973a). Disordered thought in schizophrenia. New York: Appleton-Century-Crofts. Chapman, L. J., & Chapman, J. P. (1973b). Problems in the measurement of cognitive deficit. Psychological Bulletin.2 2380-385. , Chapman L. J., & Chapman, J. P. (1978). The measurement of differential deficit. Journal of Psychiatric Research. 14,303-311. 221
Chapman, L. J., Chapman, J. P., & Miller, G. A. (1964). A theory of verbal behavior in schizophrenia. In B. A. Maher (Ed.), Progress in experimental personality research (Vol. 1, pp. 49-77). New York: Academic Press.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement. 2Q, 37-46.
Cohen, J. (1968). Weighted kappa: Nominal scale agreement with provision for scaled disagreement or partial credit. Psychological Bulletin. 20,213-220. Cohen, J. D., & Servan-Schreiber, D. (1992). Context, cortex, and dopamine: A connectionist approach to behavior and biology in schizophrenia. Psychological Review. 22»45-77.
Cramer, P. (1970). A study of homographs. In L. Postman & G. Keppel (Eds.), Norms of word association (pp. 361-382). New York: Academic Press.
Cromwell, R. L., & Dokecki, P. R. (1968). Schizophrenic language: A disattention interpretation. In S. Rosenberg & J. H. Koplin (Eds.), Developments in applied psycholinguistics research (pp. 209-260). New York: Macmillan.
Davis, J. M. (1976). Comparative doses and costs of antipsychotic medication. Archives of General Psychiatry. 33. 858-861.
Davis, J. M. (1985). Organic therapies: Antipsychotic drugs. In H. I. Kaplan & B. J. Sadock (Eds.). Comprehensive textbook of*psychiatry (4th ed., pp. 1481-1513). Baltimore: Williams & Wilkins. Deckner, C. W., Greenberg, H. C„ & Cash, T. F. (1971). The relationship of cue strength and temporal interval to schizophrenic associative disorders. Journal of JteM ogy, 22,123-130.
Dykes, M., & McGhie, A. (1976). A comparative study of attentional strategies of schizophrenic and highly creative normal subjects. British Journal of Psychiatry. 128. 50-56.
Einstein, G. 0., & Hunt, R. R. (1980). Levels of processing and organization: Additive effects of individual-item and relational processing. Journal of Experimental Psychology: Human Learning and Memory. &, 588-598.
Ervin-Tripp, S. M. (1970). Substitution, context, and association. In L. Postman & G. Keppel (Eds.), Norms of word association (pp. 383-467). New York: Academic Press.
Feinberg, I., & Mercer, M. (1960). Studies of thought disorder in schizophrenia. I. Chapman’s tests of distractibility and associative intrusion in schizophrenia and organic brain disease. A.M. A. Archives of General Psychiatry. 2,504-511.
Gonen, J. Y. (1970). Associative interference in schizophrenia as a function of paranoid status and premorbid adjustment. Journal of Consulting and Clinical Psychology. 24,221-225. 222
Gottesman, 1.1., & Shields, J. (1972). Schizophrenia and genetics: A twin study vantage point. New York: Academic Press.
Green D. M., & Swets, J. A. (1966). Signal detection theory and psvchophvsics. New York: Wiley. Grimes, C., & McGhie, A. (1973). Stimulus overload in schizophrenia. Canadian Journal of Behavioral Science. 5,101-110.
Haley, J. (1959). An interactional description of schizophrenia. Psychiatry. 22. 321- 332.
Hamsher, K. S., & Arnold, K. O. (1976). A test of Chapman’s theory of schizophrenic thought disorder. Journal of Abnormal Psychology. £2,296-302.
Harrow, M., Grossman, L. S., Silverstein, M. L., Meltzer, H. Y., & Kettering, R. L. (1986). A longitudinal study of thought disorder in manic patients. Archives of General Psychiatry. 42,781-785.
Harrow, M., & Marengo, J. T. (1986). Schizophrenic thought disorder at followup: Its persistence and prognostic significance. Schizophrenia Bulletin. 12, 373-393.
Harrow, M., Marengo, J., & McDonald, C. (1986). The early course of schizophrenic thought disorder. Schizophrenia Bulletin. 12.208-224.
Harrow, M., Silverstein, M., & Marengo, J. (1983). Disordered thinking: Does it identify nuclear schizophrenia? Archives of General Psychiatry. 4Q, 765-771.
Harvey, P. D., Docherty, N. M., Serper, M. R., & Rasmussen, M. (1990). Cognitive deficits and thought disorder: H An 8-month followup study. Schizophrenia Bulletin. 16.147-156.
Harvey, P. D., & Serper, M. R. (1990). Linguistic and cognitive failures in schizophrenia: A multivariate analysis. Journal of Nervous and Mental Disease. 178.487-493.
Harvey, P. D., Walker, E., & Wielgus, M. S. (1986). Psychological markers of vulnerability to schizophrenia: Research and future directions. In B. A. Maher & W. B. Maher (Eds.), Progress in experimental personality research (Vol. 14, pp. 231-267). Orlando: Academic Press.
Hemsley, D. R. (1976). Problems in the interpretation of cognitive abnormalities in schizophrenia. British Journal of Psychiatry, 122, 32-35.
Hemsley, D. R., & Hawks, D. V. (1974). Speed of response and associative errors in schizophrenia. British Journal of Social and Clinical Psychology. 12,293-303.
Heston, L. L. (1966). Psychiatric disorders in foster home reared children of schizophrenic mothers. British Journal of Psychiatry. U 2, 819-825.
Hollingshead, A. B. (1957). Two factor index of social position. New Haven: Author. 223
Hollister, L. E., & Csemansky, J. G. (1990). Clinical pharmacology of psychotherapeutic drugs. New York: Livingstone.
Holzman, P. S. (Ed.). (1986). Thought disorder. Schizophrenia Bulletin. 12(3).
Holzman, P. S., Shenton, M. E., & Solovay, M. R. (1986). Quality of thought disorder in differential diagnosis. Schizophrenia Bulletin. 12. 360-372.
Hunt, R. R., & Einstein, G. O. (1981). Relational and item-specific information in memory. Journal of Verbal Learning and Verbal Behavior. 2Q, 497-514. Hurt, S. W., Holzman, P. S., & Davis, J. M. (1983). Thought disorder: The measurement of its changes. Archives of General Psychiatry. 4Q, 1281-1285.
Jenkins, J. J. (1970). The 1952 Minnesota word association norms. In L. Postman & G. Keppel (Eds.), Norms of word association (pp. 1-38). New York: Academic Press.
Keppel, G., & Strand, B. Z. (1970). Free-association responses to the primary responses and other responses selected from the Palermo-Jenkins norms. In L. Postman & G. Keppel (Eds.), Norms of word association (pp. 177-239). New York: Academic Press.
Klinka, J., & Papageorgis, D. (1976). Associative intrusions in the vocabulary of schizophrenic and other patients. British Journal of Psychiatry. 129. 584-591.
Klinka, J., & Papageorgis, D. (1984). Thought disorder in hospitalized and nonhospitalized chronic schizophrenic and nonpsychiatric patients. Canadian Journal of BshavioraLSclence, 16,130-143. Koh, S. D. (1978). Remembering of verbal materials by schizophrenic young adults. In S. Schwartz (Ed.), Language and cognition in schizophrenia (pp. 55-99). Hillsdale, NJ: Erlbaum.
Koh, S. D., Kayton, L., & Peterson, R. A. (1976). Affective encoding and consequent remembering in schizophrenic young adults. Journal of Abnormal Psychology. £5,156-166.
Kopfstein, J. H., & Neale, J. M. (1972). A multivariate study of attention dysfunction in schizophrenia. Journal of Abnormal Psychology. 80.294-298.
Kraepelin, E. (1971). Dementia praecox and paraphrenia (R. M. Barclay, Trans.). Huntington, NY: Krieger. (Original work published 1919)
Kwapil, T. R., Hegley, D. C., Chapman, L. J., & Chapman, J. P. (1990). Facilitation of word recognition by semantic priming in schizophrenia. Journal of Abnormal Psychology.22, 215-221.
Larsen, S. F., & Fromholt, P. (1976). Mnemonic organization and free recall in schizophrenia. Journal of Abnormal Psychology. £5,61-65.
Macmillan, N. A., & Creelman, C. D. (1991). Detection theory: A user’s guide. New York: Cambridge University Press. 224
Maher, B. A. (1974). [Editorial]. Journal of Consulting and Clinical Psychology. 42.1- 3. Maher, B. A. (1983). A tentative theory of schizophrenic utterance. In B. A. Maher & W. B. Maher (Eds.), Progress in experimental personality research (Vol. 12, pp. 1- 52). New York: Academic Press.
Marshall, W. L. (1973). Cognitive functioning in schizophrenia. II. Conceptual performance under unpaced, speeded, and slowed conditions. British Journal of Psychiatry. 123.423-428.
Marshall, G. R., & Cofer, C. N. (1970). Single-word ffee-association norms for 328 responses from the Connecticut cultural norms for verbal items in categories. In L. Postman & G. Keppel (Eds.), Norms of word association (pp. 321-360). New York: Academic Press.
Meyer, D. E., & Schvaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence of a dependence between retrieval operations. Journal of Experimental Psychology. 20,227-234.
Miller, B. (1974). Semantic misinterpretations of ambiguous communications in schizophrenia. Archives of General Psychiatry. 20,435-440.
Mourer, S. A. (1973). A prediction of patterns of schizophrenic error resulting from semantic generalization. Journal of Abnormal Psychology. 81. 250-254.
Myers, J. L. (1979). Fundamentals of experimental design (3rd ed.). Boston: Allyn and Bacon.
Naficy, A., & Willerman, L. (1980). Excessive yielding to normal biases is not a distinctive sign of schizophrenia. Journal of Abnormal Psychology. 22,697-703.
Neale, J. M., & Oltmanns, T. F. (1980). Schizophrenia. New York: Wiley.
Neale, J. M., Oltmanns, T. F., & Harvey, P. D. (1985). The need to relate cognitive deficits to specific behavioral referents of schizophrenia. Schizophrenia Bulletin. J I, 286-291.
Neuringer, C., Fiske, J. P., Schmidt, M. W., & Goldstein, G. (1972). Adherence to strong verbal meaning definitions in schizophrenics. Journal of Genetic Esy.ch9to.gy, 1 2315-323. 1 , Neuringer, G , Kaplan, H. A., & Goldstein, G. (1974). Schizophrenic avoidance of strong meaning associations to emotional words. Journal of Genetic Psychology. 124.123-129.
Oltmanns, T. F„ Murphy, R., Berenbaum, H., & Dunlop, S. R. (1985). Rating verbal communication impairment in schizophrenia and affective disorders. Schizophrenia Bulletin. H , 292-299.
Palermo, D. S., & Jenkins, J. J. (1964). Word association norms: Grade school through college. Minneapolis: University of Minnesota. 225
Paulman, R., & Meyers-Abell, J. (1980). Chapman’s versus Broen and Storms’ theory of schizophrenic thought disorder: An empirical comparison. Journal of Clinical Psychology. 26, 844-852.
Persons, J. B„ & Baron, J. (1985). Processes underlying formal thought disorder in psychiatric inpatients. Journal of Nervous and Mental Disease. 173. 667-676.
Postman, L. (1970). The California norms: Association as a function of word frequency. In L. Postman & G. Keppel (Eds.), Norms of word association (pp. 241-320). New York: Academic Press.
Rattan, R. B., & Chapman, L. J. (1973). Associative intrusions in schizophrenic verbal behavior. Journal of Abnormal Psychology. 82.169-173.
Roberts, M. A., & Schuham, A. I. (1974). Word associations of schizophrenics and alcoholics as a function of strength of associative distracter. Journal of Abnormal Psychology. 426-431.
Rosenthal, D. (1970). Genetic theory and abnormal behavior. New York: McGraw- Hill.
Saccuzzo, D. P., & Braff, D. L. (1980). Associative cognitive dysfunction in schizophrenia and old age. Journal of Nervous and Mental Disease. 1 6 $ , 41-45.
Salzinger, K. (1973). Schizophrenia: Behavioral aspects. New York: Wiley.
Schvaneveldt, R. W., & Meyer, D. E. (1973). Retrieval and comparison processes in semantic memory. In S. Komblum (Ed.), Attention and performance (Vol. 4, pp. 395-410). New York: Academic Press.
Sengel, R. A., Lovallo, W. R., Pishkin, V., Leber, W. R., & Shaffer, B. (1984). Associative response bias and severity of thought disorder in schizophrenia and mania. Journal of Clinical Psychology. 4Q, 889-892.
Shakow, D. (1950). Some psychological features of schizophrenia. In M. Reymert (Ed.), Feelings and emotions: The Mooseheart symposium (pp. 383-390). New York: McGraw-Hill.
Solovay, M. R., Shenton, M. E„ & Holzman, P. S. (1987). Comparative studies of thought disorders: I. Mania and schizophrenia. Archives of General Psychiatry. 44,13-20.
Spitzer, R. L., Endicott, J., & Robins, E. (1984). Research diagnostic criteria (RPC) for a selected group of functional disorders (rev. 3rd ed.). New York: New York State Psychiatric Institute.
Spohn, H. E., Coyne, L., Larson, J., Mittleman, F., Spray, J., & Hayes, K. (1986). Episodic and residual thought pathology in chronic schizophrenics: Effect of neuroleptics. Schizophrenia Bulletin. 12,394-407. 226 Strauss, M. E. (1975). Strong meaning-response bias in schizophrenia. Journal of Abnormal Psychology. M , 295-298.
Swinney, D. A. (1979). Lexical access during sentence comprehension: (Reconsideration of context effects. Journal of Verbal Learning and Verbal Behavior. I£, 645-659.
Swinney, D. (1984). Theoretical and methodological issues in cognitive science: A psycholinguistic perspective. In W. Kintsch, J. R. Miller, & P. G. Poison (Eds.), Methods and tactics in cognitive science (pp. 217-233). Hillsdale, NJ: Erlbaum.
Swinney, D. A., Onifer, W., Prather, P., & Hirshkowitz, M. (1979). Semantic facilitation across sensory modalities in the processing of individual words and sentences. Memory and Cognition. 7.159-165.
Thomson, D. M., & Tulving, E. (1970). Associative encoding and retrieval: Weak and strong cues. Journal of Experimental Psychology. 86. 255-262.
Thorndike, E. L., & Lorge, I. (1944). The teacher’s word book of 30.000 words. New York: Columbia University, Teachers College.
Tulving, E., & Thomson, D. M. (1971). Retrieval processes in recognition memory: Effects of associative context. Journal of Experimental Psychology. 87.116-124.
Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review. 2Q, 352-373.
Usdansky, G., & Chapman, L. J. (1960). Schizophrenic-like responses in normal subjects under time pressure. Journal of Abnormal and Social Psychology. 60. 143-146.
Venables, P. H. (1964). Input dysfunction in schizophrenia. In B. A. Maher (Ed.), Progress in experimental personality research (Vol. 1, pp. 1-47). New York: Academic Press.
Vigotsky, L. S. (1934). Thought in schizophrenia. Archives of Neurology and Psychiatry. 31.1063-1077.
Von Domarus, E. (1944). The specific laws of logic in schizophrenia. In J. S. Kasanin (Ed.), Language and thought in schizophrenia (pp. 104-114). New York: Norton.
Watkins, M. J., & Tulving, E. (1975). Episodic memory: When recognition fails. Journal of Experimental Psychology: General. 1£4,5-29.
Watson, C. G., & Plemel, D. (1981). A test of the interference and normal associate bias theories of schizophrenic cognitive deficit. Journal of Nervous and Mental Disease, 162,185-190. Wechsler, D. (1981). Wechsler Adult Intelligence Scale-Revised. New York: Psychological Corporation. 227
Weckowicz, T. E., & Blewett, D. B. (1959). Size constancy and abstract thinking in schizophrenic patients. Journal of Mental Science. 105. 909-934.
Whitaker, L. C. (1980). Whitaker Index of Schizophrenic Thinking. Los Angeles: Western Psychological Services.
Wielgus, M. S., & Harvey, P. D. (1988). Dichotic listening and recall in schizophrenia and mania. Schizophrenia Bulletin. 14, 689-700.
Williams, R. M„ Hemsley, D. R., & Denning-Duke, C. (1976). Language behaviour in acute and chronic schizophrenia. British Journal of Social and Clinical Psychology. 15,73-83.
Winer, B. J. (1971). Statistical principles in experimental design (2nd ed.). New York: McGraw-Hill.
Zubin, J., & Spring, B. (1977). Vulnerability: A new view of schizophrenia. Journal of Abnormal Psychology. 86.103-126.